Main Page: Difference between revisions

From Essential
Jump to navigation Jump to search
Tcepo (talk | contribs)
veille GitHub 2026-11-21 : 3 ajout(s) — unsloth, LocalAI, claude-code-templates
Tcepo (talk | contribs)
No edit summary
 
(3 intermediate revisions by the same user not shown)
Line 3: Line 3:
= infocepo.com – Cloud, AI & Labs =
= infocepo.com – Cloud, AI & Labs =


Bienvenue sur le portail '''infocepo.com'''.
Welcome to the '''infocepo.com''' portal.


Ce wiki documente l’écosystème '''Cloud, IA, automatisation et lab''' d’Infocepo.   
This wiki documents the '''Cloud, AI, automation and lab''' ecosystem of Infocepo.   
Il s’adresse aux :
It is intended for:


* administrateurs systèmes,
* system administrators,
* ingénieurs cloud,
* cloud engineers,
* développeurs,
* developers,
* étudiants,
* students,
* curieux qui veulent apprendre en pratiquant.
* curious minds who want to learn by doing.


L’objectif est simple : transformer la théorie en '''scripts réutilisables, schémas, architectures, APIs et laboratoires concrets'''.
The goal is simple: transform theory into '''reusable scripts, diagrams, architectures, APIs, and hands-on labs'''.


__TOC__
__TOC__
Line 20: Line 20:
----
----


= Accès rapide =
= Quick Access =


== Portail principal ==
== Main Portal ==
* [https://infocepo.com infocepo.com]
* [https://infocepo.com infocepo.com]


== Assistant IA ==
== AI Assistant ==
* [https://chat.infocepo.com Chat assistant]
* [https://chat.infocepo.com Chat assistant]


== Liste des pages du wiki ==
== Wiki Page List ==
* [[Special:AllPages|Toutes les pages]]
* [[Special:AllPages|All Pages]]


== Vue d’ensemble ==
== Overview ==
[[File:Ailab-architecture.png|thumb|'''Infra architecture overview''']]
[[File:Ailab-architecture.png|thumb|'''Infra architecture overview''']]


= Démarrer rapidement =
= Getting Started =


== Parcours recommandés ==
== Recommended Paths ==


; 1. Construire un assistant IA privé
; 1. Build a private AI assistant
* Déployer une stack type '''Hermes WebUI + Ollama + GPU'''
* Deploy a stack like '''Hermes WebUI + Ollama + GPU'''
* Ajouter un modèle de chat et un modèle de résumé
* Add a chat model and a summarization model
* Brancher des données internes via '''RAG + embeddings'''
* Connect internal data via '''RAG + embeddings'''


; 2. Lancer un lab cloud
; 2. Launch a cloud lab
* Créer un petit cluster Kubernetes, OpenStack ou bare-metal
* Create a small Kubernetes, OpenStack, or bare-metal cluster
* Mettre en place un pipeline de déploiement (Helm, Ansible, Terraform…)
* Set up a deployment pipeline (Helm, Ansible, Terraform…)
* Ajouter un service IA : transcription, résumé, chatbot, OCR…
* Add an AI service: transcription, summarization, chatbot, OCR…


; 3. Préparer un audit ou une migration
; 3. Prepare an audit or migration
* Inventorier les serveurs avec '''ServerDiff.sh'''
* Inventory servers with '''ServerDiff.sh'''
* Concevoir l’architecture cible
* Design the target architecture
* Automatiser la migration avec des scripts reproductibles
* Automate the migration with reproducible scripts


== Vue d’ensemble du contenu ==
== Content Overview ==
* '''Guides IA & outils''' : assistants, modèles, évaluation, GPU, RAG
* '''AI Guides & Tools''': assistants, models, evaluation, GPU, RAG
* '''Cloud & infrastructure''' : Kubernetes, OpenStack, HA, HPC, DevSecOps
* '''Cloud & Infrastructure''': Kubernetes, OpenStack, HA, HPC, DevSecOps
* '''Labs & scripts''' : audit, migration, automatisation
* '''Labs & Scripts''': audit, migration, automation
* '''Comparatifs''' : Kubernetes vs OpenStack vs AWS vs bare-metal, etc.
* '''Comparisons''': Kubernetes vs OpenStack vs AWS vs bare-metal, etc.


----
----
Line 65: Line 65:
[[File:Automation-full-vs-humans.png|thumb|right|The world after automation]]
[[File:Automation-full-vs-humans.png|thumb|right|The world after automation]]


Le but à long terme est de construire un environnement où :
The long-term goal is to build an environment where:


* les assistants IA privés accélèrent la production,
* private AI assistants accelerate production,
* les tâches répétitives sont automatisées,
* repetitive tasks are automated,
* les déploiements sont industrialisés,
* deployments are industrialized,
* l’infrastructure reste '''compréhensible, portable et réutilisable'''.
* infrastructure remains '''understandable, portable, and reusable'''.


Chaque sortie IA est mesurée en temps réel, les modèles faibles sont remplacés automatiquement, et les tâches sans valeur sont coupées. L'autonomie vient de mécanismes techniques qui fonctionnent quand personne ne regarde — pas de règles écrites dans un doc.
Every AI output is measured in real time, weak models are automatically replaced, and valueless tasks are cut. Autonomy comes from technical mechanisms that work when no one is watching — not from rules written in a document.


[[File:SUMMARY-DIAGRAM-7311e6b1-aede-4989-ade2-a42d1a6e0ff2.png|thumb|right|Main page summary]]
[[File:SUMMARY-DIAGRAM-7311e6b1-aede-4989-ade2-a42d1a6e0ff2.png|thumb|right|Main page summary]]
Line 78: Line 78:
----
----


= Catalogue rapide des services =
= Quick Service Catalog =


{| class="wikitable"
{| class="wikitable"
|+ Services principaux
|+ Main Services
! Catégorie !! Service !! Rôle
! Category !! Service !! Role
|-
|-
| API || [https://api-llm-custom.ailab.infocepo.com/docs LLM] || Modèles de chat, code, RAG, OCR
|| API || [https://api-llm-custom.ailab.infocepo.com/docs LLM] || Chat, code, RAG, OCR models
|-
|-
| API || [https://api-audio2txt.ailab.infocepo.com/docs STT] || Transcription audio
|| API || [https://api-audio2txt.ailab.infocepo.com/docs STT] || Audio transcription
|-
|-
| API || [https://api-tts-omnivoice.ailab.infocepo.com/docs TTS] || Synthèse vocale
|| API || [https://api-tts-omnivoice.ailab.infocepo.com/docs TTS] || Speech synthesis
|-
|-
| API || [https://github.com/ynotopec/api-realtime-ai realtime-ai] || Temps réel WebSocket / WebRTC
|| API || [https://github.com/ynotopec/api-realtime-ai realtime-ai] || Real-time WebSocket / WebRTC
|-
|-
| API || [https://api-llm-custom.ailab.infocepo.com/docs IMAGE2TXT] || OCR / VLM via endpoint dédié
|| API || [https://api-llm-custom.ailab.infocepo.com/docs IMAGE2TXT] || OCR / VLM via dedicated endpoint
|-
|-
| API || [https://api-summary.ailab.infocepo.com:wait-2026-12/docs summary] || Résumé de textes longs
|| API || [https://api-summary.ailab.infocepo.com:wait-2026-12/docs summary] || Long text summarization
|-
|-
| API || [https://api-embedding.ailab.infocepo.com/docs EMBEDDINGS] || Embeddings pour RAG
|| API || [https://api-embedding.ailab.infocepo.com/docs EMBEDDINGS] || Embeddings for RAG
|-
|-
| API || [https://chromadb.ailab.infocepo.com:wait-2026-12 ChromaDB] || Base vecteur
|| API || [https://chromadb.ailab.infocepo.com:wait-2026-12 ChromaDB] || Vector database
|-
|-
| API || [https://api-txt2image.ailab.infocepo.com/docs TXT2IMAGE] || Génération d’images
|| API || [https://api-txt2image.ailab.infocepo.com/docs TXT2IMAGE] || Image generation
|-
|-
| API || [https://api-diarization.ailab.infocepo.com/docs diarization] || Segmentation locuteurs
|| API || [https://api-diarization.ailab.infocepo.com/docs diarization] || Speaker diarization
|-
|-
| Observabilité || [https://grafana.ailab.infocepo.com:wait-2026-12 monitoring] || Dashboards techniques
|| Observability || [https://grafana.ailab.infocepo.com:wait-2026-12 monitoring] || Technical dashboards
|-
|-
| Observabilité || [https://uptime-kuma.ailab.infocepo.com:wait-2026-12/status/ai status] || Disponibilité des services
|| Observability || [https://uptime-kuma.ailab.infocepo.com:wait-2026-12/status/ai status] || Service availability
|-
|-
| Observabilité || [https://web-stat.c1.ailab.infocepo.com:wait-2026-12 web-stat] || Statistiques web
|| Observability || [https://web-stat.c1.ailab.infocepo.com:wait-2026-12 web-stat] || Web statistics
|-
|-
| Observabilité || [https://api.ailab.infocepo.com:wait-2026-09/ui LLM-stat] || Vue API / usage
|| Observability || [https://api.ailab.infocepo.com:wait-2026-09/ui LLM-stat] || API / usage dashboard
|-
|-
| Outils || [https://datalab.ailab.infocepo.com:wait-2026-12 dataLab] || Environnement de travail hors-production
|| Tools || [https://datalab.ailab.infocepo.com:wait-2026-12 dataLab] || Non-production work environment
|-
|-
| Outils || [https://translate-rt.ailab.infocepo.com realtime translation] || Traduction
|| Tools || [https://translate-rt.ailab.infocepo.com realtime translation] || Translation
|-
|-
| Outils || [https://demos.ailab.infocepo.com Demos] || Démonstrateurs
|| Tools || [https://demos.ailab.infocepo.com Demos] || Demonstrators
|}
|}


----
----


= Sélection AI & architecture par couche =
= AI Selection & Layered Architecture =


Le tableau de bord [https://ai.arch.infocepo.com '''Sélection AI'''] présente l'architecture par couche des modèles IA choisies :
The [https://ai.arch.infocepo.com '''AI Selection'''] dashboard presents the layered architecture of selected AI models:


{| class="wikitable"
{| class="wikitable"
! Couche !! Modèle !! Alias(s) !! Taille
! Layer !! Model !! Alias(es) !! Size
|-
|-
| rowspan="7" style="text-align:center; vertical-align:middle; background:#f0f0f0;"| '''HYBRID-CLOUD'''
| rowspan="7" style="text-align:center; vertical-align:middle; background:#f0f0f0;"| '''HYBRID-CLOUD'''
| '''Qwen3.6-35B''' || flash, default, thinking, vision || Option 1, 35B MoE / 3B actif
| '''Qwen3.6-35B''' || flash, default, thinking, vision || Option 1, 35B MoE / 3B active
|-
|-
| '''Qwen3.8-Flash-Next''' || flash, default, thinking, vision, max || Option 2, 125B MoE, 51B n-grames internes, 6B actif, 1M contexte
| '''Qwen3.8-Flash-Next''' || flash, default, thinking, vision, max || Option 2, 125B MoE, 51B internal n-grams, 6B active, 1M context
|-
|-
| '''MiniMax-H3''' || video || 33B dense
| '''MiniMax-H3''' || video || 33B dense
Line 143: Line 143:
| '''GLiNER2.5-Multi-V1''' || privacy || 287M, mDeBERTa-v3-base — privacy filter: [https://api-llm-privacy-proxy-gliner2-5.infocepo.com api-llm-privacy-proxy-gliner2-5]
| '''GLiNER2.5-Multi-V1''' || privacy || 287M, mDeBERTa-v3-base — privacy filter: [https://api-llm-privacy-proxy-gliner2-5.infocepo.com api-llm-privacy-proxy-gliner2-5]
|-
|-
| '''Public-Cloud''' || '''GLM-5.3-Flash''' || max || 320B MoE / 18B actif — 1M contexte, multimodal
| '''Public-Cloud''' || '''GLM-5.3-Flash''' || max || 320B MoE / 18B active — 1M context, multimodal
|-
|-
| rowspan="2" style="text-align:center; vertical-align:middle; background:#f0f0f0;"| '''LOCAL'''
| rowspan="2" style="text-align:center; vertical-align:middle; background:#f0f0f0;"| '''LOCAL'''
Line 154: Line 154:


{| class="wikitable"
{| class="wikitable"
! Outil !! Usage
! Tool !! Usage
|-
|-
| '''Hermes''' || Usage général, raisonnement
|| '''Hermes''' || General purpose, reasoning
|-
|-
| '''OpenCode''' || DevSecOps
|| '''OpenCode''' || DevSecOps
|-
|-
| '''Kilo Code''' || Développement pur
|| '''Kilo Code''' || Pure development
|}
|}


== Mémoire ==
== Memory ==


Pour la couche d'embedding vectoriel (RAG, mémoire sémantique) :
For the vector embedding layer (RAG, semantic memory):


* '''BGE-M3''' — 568M paramètres, architecture XLM-RoBERTa. Multi-fonction (dense + sparse + multi-vector), 100+ langues, licence MIT. Idéal comme moteur de mémoire à échelle.
* '''BGE-M3''' — 568M parameters, XLM-RoBERTa architecture. Multi-purpose (dense + sparse + multi-vector), 100+ languages, MIT license. Ideal as a scale-out memory engine.


= Nouveautés =
= What's New =


=== Nouveautés implémentés ===
=== Implemented Features ===
* [https://github.com/ynotopec/audio2audio-translate-next '''audio2audio-translate-next'''] (2026-11-10) : aucune description GitHub
* [https://github.com/ynotopec/translate-rt '''translate-rt'''] (2026-11-19) : Real-time translation API — multilingual translation with low latency
* [https://github.com/ynotopec/translate-rt '''translate-rt'''] (2026-11-07) : Real-time translation API — multilingual translation with low latency
* [https://github.com/ynotopec/gouv-fr-skills '''gouv-fr-skills'''] (2026-11-19) : no GitHub description
* [https://github.com/ynotopec/audio2audio-translate-next '''audio2audio-translate-next'''] (2026-11-17) : no GitHub description
* [https://github.com/ynotopec/api-rag '''api-rag'''] (2026-11-16) : RAG API — semantic search over knowledge bases with retrieval-augmented generation
* [https://github.com/ynotopec/api-txt2image '''api-txt2image'''] (2026-11-16) : Text-to-image generation API for AI-powered image creation
* [https://github.com/ynotopec/diff-multi '''diff-multi'''] (2026-11-07) : Multi-format diff tool — compare files, directories, and structured data
* [https://github.com/ynotopec/diff-multi '''diff-multi'''] (2026-11-07) : Multi-format diff tool — compare files, directories, and structured data
* [https://github.com/ynotopec/gpu-cluster '''gpu-cluster'''] (2026-11-07) : GPU cluster management — k8s-based AI compute cluster with NVIDIA GPUs
* [https://github.com/ynotopec/gpu-cluster '''gpu-cluster'''] (2026-11-07) : GPU cluster management — k8s-based AI compute cluster with NVIDIA GPUs
Line 190: Line 193:
* [https://github.com/ynotopec/legal-agent '''legal-agent'''] (2026-11-07) : Legal AI agent — assist with legal document analysis and research
* [https://github.com/ynotopec/legal-agent '''legal-agent'''] (2026-11-07) : Legal AI agent — assist with legal document analysis and research
* [https://github.com/ynotopec/image-editor '''image-editor'''] (2026-11-07) : AI image editor — modify, enhance, and generate images with AI
* [https://github.com/ynotopec/image-editor '''image-editor'''] (2026-11-07) : AI image editor — modify, enhance, and generate images with AI
* [https://github.com/ynotopec/api-rag '''api-rag'''] (2026-11-07) : RAG API — semantic search over knowledge bases with retrieval-augmented generation
* [https://github.com/ynotopec/sam-audio '''sam-audio'''] (2026-11-07) : ai audio processing
* [https://github.com/ynotopec/sam-audio '''sam-audio'''] (2026-11-07) : ai audio processing
* [https://github.com/ynotopec/ai-security '''ai-security'''] (2026-11-07) : AI security tools — audit and harden LLM deployments against common attacks
* [https://github.com/ynotopec/ai-security '''ai-security'''] (2026-11-07) : AI security tools — audit and harden LLM deployments against common attacks
* [https://github.com/ynotopec/artificial-human '''artificial-human'''] (2026-11-07) : Artificial human simulation — explore the boundary between human and AI behavior
* [https://github.com/ynotopec/artificial-human '''artificial-human'''] (2026-11-07) : Artificial human simulation — explore the boundary between human and AI behavior
* [https://github.com/ynotopec/mediawiki-migrate '''mediawiki-migrate'''] (2026-11-07) : MediaWiki migration tool — migrate and sync wiki content between platforms
* [https://github.com/ynotopec/mediawiki-migrate '''mediawiki-migrate'''] (2026-11-07) : MediaWiki migration tool — migrate and sync wiki content between platforms
* [https://github.com/ynotopec/ai-assistant '''ai-assistant'''] (2026-11-07) : General-purpose AI assistant with task execution capabilities
=== Top Tasks — Deployed Infrastructure ===
* [https://github.com/ynotopec/opencode-demo '''opencode-demo'''] (2026-11-07) : OpenCode CLI demo — open-source alternative for AI-assisted coding
''Generated from real infrastructure (kubectl + LLM endpoints) on 2026-10-03. Do not edit manually — regenerate after deployment.''
=== Top tasks — infrastructure déployée ===
''Généré depuis l'infra réelle (kubectl + endpoints LLM) le 2026-10-03. Ne pas éditer à la main — regénérer après déploiement.''


==== Endpoints LLM (vérifiés) ====
==== LLM Endpoints (verified) ====
{| class="wikitable"
{| class="wikitable"
! Endpoint !! Modèle réel !! URL
! Endpoint !! Real Model !! URL
|-
|-
| '''ai-thinking''' || qwen3.6 (thinking) || <code>https://api-llm-custom.ailab.infocepo.com</code>
|| '''ai-thinking''' || qwen3.6 (thinking) || <code>https://api-llm-custom.ailab.infocepo.com</code>
|-
|-
| '''ai-default''' || qwen3.6 (défaut) || <code>https://api-llm-custom.ailab.infocepo.com</code>
|| '''ai-default''' || qwen3.6 (default) || <code>https://api-llm-custom.ailab.infocepo.com</code>
|-
|-
| '''ai-flash''' || qwen3.6 (fast) || <code>https://api-llm-custom.ailab.infocepo.com</code>
|| '''ai-flash''' || qwen3.6 (fast) || <code>https://api-llm-custom.ailab.infocepo.com</code>
|-
|-
| '''ai-vision''' || qwen3.6 (vision/OCR) || <code>https://api-llm-custom.ailab.infocepo.com</code>
|| '''ai-vision''' || qwen3.6 (vision/OCR) || <code>https://api-llm-custom.ailab.infocepo.com</code>
|-
|-
| '''ai-embedding''' || bge-m3 || <code>https://api-embedding.ailab.infocepo.com</code>
|| '''ai-embedding''' || bge-m3 || <code>https://api-embedding.ailab.infocepo.com</code>
|}
|}


==== Services & infrastructure IA ====
==== AI Services & Infrastructure ====
* [https://agents-saas.ailab.infocepo.com '''Agents SaaS'''] : Déploiement multi-tenant d'agents Hermes
* [https://agents-saas.ailab.infocepo.com '''Agents SaaS'''] : Multi-tenant deployment of Hermes agents
* [https://bge-m3-article.ailab.infocepo.com '''bge-m3 Article'''] : Recherche sémantique sur articles (bge-m3)
* [https://bge-m3-article.ailab.infocepo.com '''bge-m3 Article'''] : Semantic search on articles (bge-m3)
* [https://infra2.ailab.infocepo.com '''Infra Dashboard v2'''] : Tableau de bord infrastructure
* [https://infra2.ailab.infocepo.com '''Infra Dashboard v2'''] : Infrastructure dashboard
* [https://keycloak.demo1.ailab.infocepo.com '''Keycloak'''] : SSO / authentification centralisée
* [https://keycloak.demo1.ailab.infocepo.com '''Keycloak'''] : SSO / centralized authentication
* [https://mcp-proxy.ailab.infocepo.com '''MCP Proxy'''] : Passerelle MCP
* [https://mcp-proxy.ailab.infocepo.com '''MCP Proxy'''] : MCP gateway
* [https://onyxia.demo1.ailab.infocepo.com '''Onyxia'''] : Data science platform (catalogue de services)
* [https://onyxia.demo1.ailab.infocepo.com '''Onyxia'''] : Data science platform (service catalog)
* [https://rag.ailab.infocepo.com '''RAG OpenWebUI'''] : RAG intégré à Open WebUI
* [https://rag.ailab.infocepo.com '''RAG OpenWebUI'''] : RAG integrated with Open WebUI
* [https://registry.demo1.ailab.infocepo.com '''Registry'''] : Registre d'images Docker local
* [https://registry.demo1.ailab.infocepo.com '''Registry'''] : Local Docker image registry
* [https://tika.demo1.ailab.infocepo.com '''Tika'''] : Extraction de texte (Apache Tika)
* [https://tika.demo1.ailab.infocepo.com '''Tika'''] : Text extraction (Apache Tika)


==== Benchmarks & évaluation ====
==== Benchmarks & Evaluation ====
* [https://benchmark.demo1.ailab.infocepo.com '''Benchmark Endpoints'''] : Comparatif de performance d'endpoints LLM
* [https://benchmark.demo1.ailab.infocepo.com '''Benchmark Endpoints'''] : LLM endpoint performance comparison
* [https://eval.ailab.infocepo.com '''Eval Dashboard'''] : Évaluation de modèles LLM
* [https://eval.ailab.infocepo.com '''Eval Dashboard'''] : LLM model evaluation
* [https://qpr.infocepo.com '''OpenRouter Q/P'''] : Classement qualité/prix des modèles OpenRouter
* [https://qpr.infocepo.com '''OpenRouter Q/P'''] : OpenRouter model quality/price ranking
* [https://oss-efficiency.ailab.infocepo.com '''OSS Efficiency'''] : Efficacité des modèles open-source
* [https://oss-efficiency.ailab.infocepo.com '''OSS Efficiency'''] : Open-source model efficiency
* [https://privacy.ailab.infocepo.com '''Privacy Benchmark'''] : Benchmark de fuites de secrets (proxy LLM)
* [https://privacy.ailab.infocepo.com '''Privacy Benchmark'''] : Secret leak benchmark (LLM proxy)
* [https://rag-bench.demo1.ailab.infocepo.com '''RAG Bench'''] : Benchmark de pipelines RAG
* [https://rag-bench.demo1.ailab.infocepo.com '''RAG Bench'''] : RAG pipeline benchmark
* [https://rag-eval.demo1.ailab.infocepo.com '''RAG Eval'''] : Évaluation qualité RAG
* [https://rag-eval.demo1.ailab.infocepo.com '''RAG Eval'''] : RAG quality evaluation


==== Veille automatisée ====
==== Automated Monitoring ====
* [https://models.ailab.infocepo.com '''Model Tracker'''] : Suivi des modèles HuggingFace vs déploiements
* [https://models.ailab.infocepo.com '''Model Tracker'''] : HuggingFace models vs deployments tracking
* [https://news-aggregator.demo1.ailab.infocepo.com '''News Aggregator'''] : Agrégation de flux d'actualité IA
* [https://news-aggregator.demo1.ailab.infocepo.com '''News Aggregator'''] : AI news feed aggregation


=== Veille GitHub (populaire) ===
=== GitHub Monitoring (Popular) ===
* [https://github.com/BerriAI/litellm '''Litellm'''] (2026-10-26) : AI gateway — appelle 100+ LLM APIs via une seule interface (62k⭐)
* [https://github.com/unslothai/unsloth '''unsloth'''] (2026-11-21) : Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more. (79k⭐)
* [https://github.com/sgl-project/sglang '''SGLang'''] (2026-10-26) : Framework de serving haute performance pour les LLMs (38k⭐)
* [https://github.com/mudler/LocalAI '''LocalAI'''] (2026-11-21) : LocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required. (50k⭐)
* [https://github.com/langchain-ai/langgraph '''LangGraph'''] (2026-10-26) : Construire des agents IA résilients et modulaires (44k⭐)
* [https://github.com/davila7/claude-code-templates '''claude-code-templates'''] (2026-11-21) : Open-source registry and CLI for Claude Code components: agents, commands, skills, hooks, MCPs, settings and mods (34k⭐)
* [https://github.com/morluto/rea '''rea'''] (2026-11-11) : Reverse engineer anything with agents, from app behavior down to native binaries. (449k⭐)
* [https://github.com/vllm-project/vllm '''vllm'''] (2026-11-20) : A high-throughput and memory-efficient inference and serving engine for LLMs (97k⭐)
* [https://github.com/NousResearch/hermes-agent '''hermes-agent'''] (2026-11-11) : The agent that grows with you (262k⭐)
* [https://github.com/ifixai-ai/iFixAi '''iFixAi'''] (2026-11-20) : Independent Auditing of AI Agents. Run by human or the agent itself, to answer the most crucial question in the AI Agent Economy. Is the ag… (46k⭐)
* [https://github.com/n8n-io/n8n '''n8n'''] (2026-11-11) : Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ inte… (210k⭐)
* [https://github.com/continuedev/continue '''continue'''] (2026-11-20) : open-source coding agent (36k⭐)
* [https://github.com/langflow-ai/langflow '''langflow'''] (2026-11-11) : Langflow is a powerful tool for building and deploying AI-powered agents and workflows. (156k⭐)
* [https://github.com/PostHog/posthog '''posthog'''] (2026-11-19) : :hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session r… (40k⭐)
* [https://github.com/stablyai/orca '''orca'''] (2026-11-11) : Orca lets you run a fleet of coding agents (e.g. Claude Code, Codex) with your own subscription. Available on desktop, mobile and remote ru… (121k⭐)
* [https://github.com/yetone/magpie '''magpie'''] (2026-11-19) : Every agent's model. One place. Codex on DeepSeek, Claude Code on Kimi, from the menu bar. (38k⭐)
* [https://github.com/earendil-works/pi '''pi'''] (2026-11-11) : AI agent toolkit: unified LLM API, agent loop, TUI, coding agent CLI (121k⭐)
* [https://github.com/onyx-dot-app/onyx '''onyx'''] (2026-11-19) : Open Source AI Platform - AI Chat with advanced features that works with every LLM (33k⭐)
* [https://github.com/anomalyco/opencode '''opencode'''] (2026-11-11) : The open source coding agent. (225k⭐)
* [https://github.com/aaif-goose/goose '''goose'''] (2026-11-18) : an open source, extensible AI agent that goes beyond code suggestions - install, execute, edit, and test with any LLM (56k⭐)
* [https://github.com/paperclipai/paperclip '''paperclip'''] (2026-11-11) : The open-source app everyone uses to manage agents at work (112k⭐)
* [https://github.com/multica-ai/multica '''multica'''] (2026-11-18) : Make humans and AI agents work as one team — open-source and self-hostable. (56k⭐)
* [https://github.com/ggml-org/llama.cpp '''llama.cpp'''] (2026-11-12) : LLM inference in C/C++ (134k⭐)
* [https://github.com/CherryHQ/cherry-studio '''cherry-studio'''] (2026-11-18) : Your AI workspace for models, agents, and productivity. (53k⭐)
* [https://github.com/thedotmack/claude-mem '''claude-mem'''] (2026-11-12) : Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and inject… (116k⭐)
* [https://github.com/firecrawl/firecrawl '''firecrawl'''] (2026-11-17) : Supercharge your AI agents with data from the web and beyond. Building the library for superintelligence. 🔥 (202k⭐)
* [https://github.com/nexu-io/open-design '''open-design'''] (2026-11-12) : The 1st Collaborative AI Design Agent Workspace. Local-first and open source. Create motion, prototypes, websites, apps, dashboards, presen… (107k⭐)
* [https://github.com/tt-a1i/archify '''archify'''] (2026-11-17) : Turn any idea, plan, or codebase into a beautiful interactive diagram. An agent skill for Claude Code, Codex, and more. (97k⭐)
* [https://github.com/Significant-Gravitas/AutoGPT '''AutoGPT'''] (2026-11-13) : AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on… (188k⭐)
* [https://github.com/vercel-labs/skills '''skills'''] (2026-11-17) : The open agent skills tool - npx skills (39k⭐)
* [https://github.com/supabase/supabase '''supabase'''] (2026-11-13) : The Postgres development platform. Supabase gives you a dedicated Postgres database to build your web, mobile, and AI applications. (114k⭐)
* [https://github.com/browser-use/browser-use '''browser-use'''] (2026-11-16) : Open-source Python library for AI browser agents, plus Browser Use Cloud: stealth browsers and a hosted agent API. (121k⭐)
* [https://github.com/cheaper-inference/OmniRoute '''OmniRoute'''] (2026-11-13) : Never stop coding. Free MIT AI gateway: one endpoint, 375 providers (150+ free), 1,400+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, Mi… (86k⭐)
* [https://github.com/rtk-ai/rtk '''rtk'''] (2026-11-16) : CLI proxy that reduces LLM token consumption by 60-90% on common dev commands. Single Rust binary, zero dependencies (86k⭐)
* [https://github.com/pbakaus/impeccable '''impeccable'''] (2026-11-15) : The design vocabulary for your coding agent. Steer with polish and bolder, review with critique and audit, catch AI slop before it ships. (110k⭐)
* [https://github.com/ChromeDevTools/chrome-devtools-mcp '''chrome-devtools-mcp'''] (2026-11-15) : Chrome DevTools for coding agents (56k⭐)
* [https://github.com/PDFMathTranslate/PDFMathTranslate '''PDFMathTranslate'''] (2026-11-15) : (EMNLP 2025 Demo) PDF scientific paper translation with preserved formats - 基于 AI 完整保留排版的 PDF 文档全文双语翻译,支持 Google/DeepL/Ollama/OpenAI 等服务,提供… (38k⭐)
* [https://github.com/openai/codex '''codex'''] (2026-11-14) : Lightweight coding agent that runs in your terminal (135k⭐)
* [https://github.com/openai/codex '''codex'''] (2026-11-14) : Lightweight coding agent that runs in your terminal (135k⭐)
* [https://github.com/heygen-com/hyperframes '''hyperframes'''] (2026-11-14) : Write HTML. Render video. Built for agents. (83k⭐)
* [https://github.com/heygen-com/hyperframes '''hyperframes'''] (2026-11-14) : Write HTML. Render video. Built for agents. (83k⭐)
* [https://github.com/vercel-labs/agent-browser '''agent-browser'''] (2026-11-14) : Browser automation CLI for AI agents (50k⭐)
* [https://github.com/vercel-labs/agent-browser '''agent-browser'''] (2026-11-14) : Browser automation CLI for AI agents (50k⭐)
* [https://github.com/pbakaus/impeccable '''impeccable'''] (2026-11-15) : The design vocabulary for your coding agent. Steer with polish and bolder, review with critique and audit, catch AI slop before it ships. (110k⭐)
* [https://github.com/Significant-Gravitas/AutoGPT '''AutoGPT'''] (2026-11-13) : AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on… (188k⭐)
* [https://github.com/ChromeDevTools/chrome-devtools-mcp '''chrome-devtools-mcp'''] (2026-11-15) : Chrome DevTools for coding agents (56k⭐)
* [https://github.com/supabase/supabase '''supabase'''] (2026-11-13) : The Postgres development platform. Supabase gives you a dedicated Postgres database to build your web, mobile, and AI applications. (114k⭐)
* [https://github.com/PDFMathTranslate/PDFMathTranslate '''PDFMathTranslate'''] (2026-11-15) : (EMNLP 2025 Demo) PDF scientific paper translation with preserved formats - 基于 AI 完整保留排版的 PDF 文档全文双语翻译,支持 Google/DeepL/Ollama/OpenAI 等服务,提供… (38k⭐)
=== Backlog / Technology Monitoring ===
* [https://github.com/huggingface/transformers '''transformers'''] (2026-11-16) : 🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models,… (169k⭐)
* [https://github.com/browser-use/browser-use '''browser-use'''] (2026-11-16) : Open-source Python library for AI browser agents, plus Browser Use Cloud: stealth browsers and a hosted agent API. (121k⭐)
* [https://github.com/rtk-ai/rtk '''rtk'''] (2026-11-16) : CLI proxy that reduces LLM token consumption by 60-90% on common dev commands. Single Rust binary, zero dependencies (86k⭐)
* [https://github.com/firecrawl/firecrawl '''firecrawl'''] (2026-11-17) : Supercharge your AI agents with data from the web and beyond. Building the library for superintelligence. 🔥 (202k⭐)
* [https://github.com/tt-a1i/archify '''archify'''] (2026-11-17) : Turn any idea, plan, or codebase into a beautiful interactive diagram. An agent skill for Claude Code, Codex, and more. (97k⭐)
* [https://github.com/vercel-labs/skills '''skills'''] (2026-11-17) : The open agent skills tool - npx skills (39k⭐)
* [https://github.com/aaif-goose/goose '''goose'''] (2026-11-18) : an open source, extensible AI agent that goes beyond code suggestions - install, execute, edit, and test with any LLM (56k⭐)
* [https://github.com/multica-ai/multica '''multica'''] (2026-11-18) : Make humans and AI agents work as one team — open-source and self-hostable. (56k⭐)
* [https://github.com/CherryHQ/cherry-studio '''cherry-studio'''] (2026-11-18) : Your AI workspace for models, agents, and productivity. (53k⭐)
* [https://github.com/PostHog/posthog '''posthog'''] (2026-11-19) : :hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session r… (40k⭐)
* [https://github.com/yetone/magpie '''magpie'''] (2026-11-19) : Every agent's model. One place. Codex on DeepSeek, Claude Code on Kimi, from the menu bar. (38k⭐)
* [https://github.com/onyx-dot-app/onyx '''onyx'''] (2026-11-19) : Open Source AI Platform - AI Chat with advanced features that works with every LLM (33k⭐)
* [https://github.com/vllm-project/vllm '''vllm'''] (2026-11-20) : A high-throughput and memory-efficient inference and serving engine for LLMs (97k⭐)
* [https://github.com/ifixai-ai/iFixAi '''iFixAi'''] (2026-11-20) : Independent Auditing of AI Agents. Run by human or the agent itself, to answer the most crucial question in the AI Agent Economy. Is the ag… (46k⭐)
* [https://github.com/continuedev/continue '''continue'''] (2026-11-20) : open-source coding agent (36k⭐)
* [https://github.com/unslothai/unsloth '''unsloth'''] (2026-11-21) : Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more. (79k⭐)
* [https://github.com/mudler/LocalAI '''LocalAI'''] (2026-11-21) : LocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required. (50k⭐)
* [https://github.com/davila7/claude-code-templates '''claude-code-templates'''] (2026-11-21) : Open-source registry and CLI for Claude Code components: agents, commands, skills, hooks, MCPs, settings and mods (34k⭐)
=== Backlog / Veille Technologique ===


=== Agents IA & Orchestration ===
=== AI Agents & Orchestration ===
* [https://github.com/langchain-ai/langchain '''LangChain'''] : Framework pour applications basées sur les LLM. Le plus mature. (145k⭐).
* [https://github.com/langchain-ai/langchain '''LangChain'''] : Framework for LLM-based applications. Most mature. (145k⭐).
* [https://github.com/langgenius/dify '''Dify'''] : Plateforme de développement d'applications IA (LLM Ops). Déployable en self-host. (154k⭐).
* [https://github.com/langgenius/dify '''Dify'''] : AI application development platform (LLM Ops). Self-hostable. (154k⭐).
* [https://github.com/open-webui/open-webui '''open-webui'''] : Interface web IA (Ollama, OpenAI API, MCP). (151k⭐).
* [https://github.com/open-webui/open-webui '''open-webui'''] : AI web interface (Ollama, OpenAI API, MCP). (151k⭐).
* [https://github.com/ollama/ollama '''ollama'''] : Get up and running with Kimi-K2.6, GLM-5.2, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models. (180k⭐).
* [https://github.com/ollama/ollama '''ollama'''] : Get up and running with Kimi-K2.6, GLM-5.2, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models. (180k⭐).
* [https://github.com/milvus-io/milvus '''milvus'''] : Base de données vectorielle haute performance. (46k⭐).
* [https://github.com/milvus-io/milvus '''milvus'''] : High-performance vector database. (46k⭐).
=== Audio & TTS ===
=== Audio & TTS ===
* [https://huggingface.co/Qwen/Qwen3-Omni-30B-A3B-Instruct '''Qwen3-Omni-30B-A3B-Instruct'''] — Modèle multimodal Qwen (audio + texte + image).
* [https://huggingface.co/Qwen/Qwen3-Omni-30B-A3B-Instruct '''Qwen3-Omni-30B-A3B-Instruct'''] — Qwen multimodal model (audio + text + image).


=== Génération & Édition d'Images ===
=== Image Generation & Editing ===
* [https://github.com/Comfy-Org/ComfyUI '''ComfyUI''' ] : GUI et backend diffusion model le plus modulaire. (131k⭐)
* [https://github.com/Comfy-Org/ComfyUI '''ComfyUI''' ] : Most modular diffusion model GUI and backend. (131k⭐)


=== RAG & Traitement de Documents ===
=== RAG & Document Processing ===


* [https://github.com/paperless-ngx/paperless-ngx '''paperless-ngx'''] : Document management : scan, index, archive. ML, OCR. (45k⭐).
* [https://github.com/paperless-ngx/paperless-ngx '''paperless-ngx'''] : Document management: scan, index, archive. ML, OCR. (45k⭐).
* [https://github.com/graphify-labs/graphify '''graphify'''] : Transforme codebase en graphe de connaissances interrogeable. (115k⭐).
* [https://github.com/graphify-labs/graphify '''graphify'''] : Transforms codebase into an queryable knowledge graph. (115k⭐).
* [https://github.com/HKUDS/LightRAG '''LightRAG'''] : RAG simple et rapide. (39k⭐).
* [https://github.com/HKUDS/LightRAG '''LightRAG'''] : Simple and fast RAG. (39k⭐).
* [https://github.com/microsoft/graphrag '''graphrag'''] : RAG modulaire basé sur les graphes. (35k⭐).
* [https://github.com/microsoft/graphrag '''graphrag'''] : Graph-based modular RAG. (35k⭐).
* [https://github.com/headroomlabs-ai/headroom '''headroom'''] : Compress tool outputs, logs, files et RAG chunks. 20-95% token savings. (69k⭐).
* [https://github.com/headroomlabs-ai/headroom '''headroom'''] : Compress tool outputs, logs, files, and RAG chunks. 20-95% token savings. (69k⭐).
* [https://github.com/mem0ai/mem0 '''Mem0'''] : — Mémorie à long terme pour agents IA. (64k⭐).
* [https://github.com/mem0ai/mem0 '''Mem0'''] : — Long-term memory for AI agents. (64k⭐).


=== APIs à Développer ===
=== APIs to Develop ===
* '''Classificateur IA''' — Classification de contenu.
* '''AI Classifier''' — Content classification.
* '''Résumé mutualisé''' — API de résumé de texte partagée.
* '''Shared Summarizer''' : Shared text summarization API.
* '''NER''' — Reconnaissance d'entités nommées.
* '''NER''' : Named Entity Recognition.
* '''Compressor''' — Compression de contenu.
* '''Compressor''' — Content compression.


=== Infrastructure & Backend ===
=== Infrastructure & Backend ===
* [https://github.com/vllm-project/semantic-router '''Semantic Router'''] — Routage sémantique de requêtes vLLM.
* [https://github.com/vllm-project/semantic-router '''Semantic Router'''] — Semantic routing for vLLM requests.


=== Outils Dev ===
=== Dev Tools ===
* [https://github.com/lutzroeder/netron '''netron'''] (2026-09-30) : Visualizer for neural network, deep learning and machine learning models (33k stars)
* [https://github.com/lutzroeder/netron '''netron'''] (2026-09-30) : Visualizer for neural network, deep learning and machine learning models (33k stars)
* [https://github.com/BasedHardware/omi '''omi'''] (2026-09-30) : AI that sees your screen, listens to your conversations and tells you what to do (13k stars)
* [https://github.com/BasedHardware/omi '''omi'''] (2026-09-30) : AI that sees your screen, listens to your conversations and tells you what to do (13k stars)
----
----


= Assistants IA & outils cloud =
= AI Assistants & Cloud Tools =


== Assistants IA ==
== AI Assistants ==


; '''Assistants IA auto-hébergés'''
; '''Self-hosted AI Assistants'''
* [https://github.com/nesquena/hermes-webui '''Hermes WebUI'''] + [https://ollama.com '''Ollama'''] + GPU   
* [https://github.com/nesquena/hermes-webui '''Hermes WebUI'''] + [https://ollama.com '''Ollama'''] + GPU   
: Stack typique pour assistant privé, API OpenAI-compatible et expérimentation locale.
: Typical stack for a private assistant, OpenAI-compatible API, and local experimentation.


== Développement, modèles & veille ==
== Development, Models & Monitoring ==


; '''Découverte de modèles'''
; '''Model Discovery'''
* [https://huggingface.co/models '''Models Trending''']
* [https://huggingface.co/models '''Models Trending''']


; '''Évaluation & benchmarks'''
; '''Evaluation & Benchmarks'''
* [https://arena.ai/leaderboard/code '''Agentic Evaluation''']
* [https://arena.ai/leaderboard/code '''Agentic Evaluation''']


; '''Outils de développement & fine-tuning'''
; '''Development & Fine-tuning Tools'''
* [https://github.com/trending?since=monthly '''Project Trending''']
* [https://github.com/trending?since=monthly '''Project Trending''']
* [https://grok.com '''News search''']
* [https://grok.com '''News search''']


== Matériel IA & GPU ==
== AI Hardware & GPU ==
* NVIDIA GH200
* NVIDIA GH200
* DGX Spark
* DGX Spark
Line 344: Line 328:
----
----


= API Realtime AI (DEV) =
= Realtime AI API (DEV) =


'''Statut :''' environnement DEV, remplaçante prévue de l’API OpenAI pour les cas temps réel.
'''Status''': DEV environment, planned replacement for the OpenAI API for real-time use cases.


== Configuration ==
== Configuration ==
{| class="wikitable"
{| class="wikitable"
! Variable !! Valeur
! Variable !! Value
|-
|-
| OPENAI_API_BASE || <code>wss://api-realtime-ai.ailab.infocepo.com:wait-2026-12/v1</code>
|| OPENAI_API_BASE || <code>wss://api-realtime-ai.ailab.infocepo.com:wait-2026-12/v1</code>
|-
|-
| OPENAI_API_KEY || <code>sk-XXXXX</code>
|| OPENAI_API_KEY || <code>sk-XXXXX</code>
|}
|}


== Dépôt GitHub ==
== GitHub Repository ==


== Page de test ==
== Test Page ==
* <code>external-test/half-duplex.html</code> — annulation d’écho + mode half-duplex.
* <code>external-test/half-duplex.html</code> — echo cancellation + half-duplex mode.


== Compatibilité ==
== Compatibility ==
Remplacer l’URL OpenAI par <code>$OPENAI_API_BASE</code> pour tester compatibilité et performances.
Replace the OpenAI URL with <code>$OPENAI_API_BASE</code> to test compatibility and performance.


----
----


= API LLM (OpenAI compatible) =
= LLM API (OpenAI Compatible) =


* URL de base : <code>https://api-llm-custom.ailab.infocepo.com/v1</code>
* Base URL: <code>https://api-llm-custom.ailab.infocepo.com/v1</code>
* Création du token : [https://llm-token.ailab.infocepo.com:wait-2026-12 OPENAI_API_KEY]
* Token creation: [https://llm-token.ailab.infocepo.com:wait-2026-12 OPENAI_API_KEY]
* Documentation : [https://api-llm-custom.ailab.infocepo.com/docs Documentation API]
* Documentation: [https://api-llm-custom.ailab.infocepo.com/docs API Documentation]


== Liste des modèles ==
== Model List ==
<pre>
<pre>
curl -X GET \
curl -X GET \
   'https://api-llm-custom.ailab.infocepo.com/v1/models' \
   'https://api-llm-custom.ailab.infocepo.com/v1/models' \
   -H 'Authorization: Bearer sk-XXXXX' \
   -H 'Authorization: Bearer ***' \
   -H 'accept: application/json' \
   -H 'accept: application/json' \
   | jq | sed -rn 's#^.*id.*: "(.*)".*$#* \1#p' | sort -u
   | jq | sed -rn 's#^.*id.*: "(.*)".*$#* \1#p' | sort -u
</pre>
</pre>


== Modèles ouverts & endpoints internes ==
== Open Models & Internal Endpoints ==


''Dernière mise à jour : 2026-10-03''
''Last updated: 2026-10-03''


Les modèles ci-dessous correspondent à des '''endpoints logiques''' exposés derrière une passerelle.
The models below correspond to '''logical endpoints''' exposed behind a gateway.


{| class="wikitable"
{| class="wikitable"
! Endpoint !! Description / usage principal
! Endpoint !! Description / Main Usage
|-
|-
| '''ai-thinking''' || '''qwen3.6 fp8''' – thinking
|| '''ai-thinking''' || '''qwen3.6 fp8''' – thinking
|-
|-
| '''ai-fast''' || '''qwen3.6 fp8''' en mode '''fast''' – vision/OCR/ai-default
|| '''ai-fast''' || '''qwen3.6 fp8''' in '''fast''' mode – vision/OCR/ai-default
|-
|-
| '''ai-embedding''' || '''bge-m3''' – recherche sémantique
|| '''ai-embedding''' || '''bge-m3''' – semantic search
|-
|-
| '''ai-stt-next''' || '''nemotron-3.5-asr-streaming-0.6b''' – streaming transcription vocale multilingual
|| '''ai-stt-next''' || '''nemotron-3.5-asr-streaming-0.6b''' – multilingual streaming voice transcription
|-
|-
| '''ai-stt''' || '''whisper3-turbo''' – transcription vocale multilingual
|| '''ai-stt''' || '''whisper3-turbo''' – multilingual voice transcription
|-
|-
| '''ai-tts''' || '''OmniVoice''' – TTS multilingual
|| '''ai-tts''' || '''OmniVoice''' – multilingual TTS
|-
|-
| '''ai-image''' || '''FLUX.2-klein''' – generate and edit image
|| '''ai-image''' || '''FLUX.2-klein''' – generate and edit image
|}
|}


== Exemple bash ==
== Bash Example ==
<pre>
<pre>
export OPENAI_API_MODEL="ai-default"
export OPENAI_API_MODEL="ai-default"
Line 412: Line 396:
export OPENAI_API_KEY="sk-XXXXX"
export OPENAI_API_KEY="sk-XXXXX"


promptValue="Quel est ton nom ?"
promptValue="What is your name?"
jsonValue='{
jsonValue='{
   "model": "'${OPENAI_API_MODEL}'",
   "model": "'${OPENAI_API_MODEL}'",
Line 421: Line 405:
curl ${OPENAI_API_BASE}/chat/completions \
curl ${OPENAI_API_BASE}/chat/completions \
   -H "Content-Type: application/json" \
   -H "Content-Type: application/json" \
   -H "Authorization: Bearer $OPENAI_API_KEY" \
   -H "Authorization: Bearer ***" \
   -d "${jsonValue}" 2>/dev/null | jq '.choices[0].message.content'
   -d "${jsonValue}" 2>/dev/null | jq '.choices[0].message.content'
</pre>
</pre>


== Vue infra LLM ==
== LLM Infra View ==
[[File:Litellm-proxy-mermaid-diagram-2024-03-24-205202.png|thumb|right]]
[[File:Litellm-proxy-mermaid-diagram-2024-03-24-205202.png|thumb|right]]


'''DEV (au choix)'''
'''DEV (at choice)'''
* '''A.''' <code>LiteLLM → vLLM/SgLang</code> : tests perf / compatibilité
* '''A.''' <code>LiteLLM → vLLM/SgLang</code> : performance/compatibility tests
* '''B.''' <code>LiteLLM → Ollama</code> : simple, rapide à itérer
* '''B.''' <code>LiteLLM → Ollama</code> : simple, fast to iterate
* '''C.''' <code>Ollama</code> direct : POC ultra-léger
* '''C.''' <code>Ollama</code> direct : ultra-lightweight POC


'''DEV – modèle FR / résumé'''
'''DEV – FR model / summarization'''
* <code>LiteLLM → Ollama /v1</code>
* <code>LiteLLM → Ollama /v1</code>


'''PROD'''
'''PROD'''
* '''Standard :''' <code>LiteLLM → vLLM/SgLang</code>
* '''Standard:''' <code>LiteLLM → vLLM/SgLang</code>
* '''Pont DEV→PROD :''' <code>LiteLLM (DEV) → LiteLLM (PROD) → vLLM/SgLang</code>
* '''DEV→PROD Bridge:''' <code>LiteLLM (DEV) → LiteLLM (PROD) → vLLM/SgLang</code>


'''Notes :'''
'''Notes:'''
* '''LiteLLM''' = passerelle unique (clés, quotas, logs)
* '''LiteLLM''' = single gateway (keys, quotas, logs)
* '''vLLM/SgLang''' = performance / stabilité en charge
* '''vLLM/SgLang''' = performance / stability under load
* '''Ollama''' = simplicité de prototypage
* '''Ollama''' = prototyping simplicity


----
----
Line 449: Line 433:
= API Image to Text =
= API Image to Text =


* Utilise l’API LLM avec un endpoint adapté à l’OCR / VLM.
* Uses the LLM API with an endpoint suited for OCR / VLM.
* Modèle recommandé : <code>ai-vision</code>
* Recommended model: <code>ai-vision</code>


== Exemple bash ==
== Bash Example ==
<pre>
<pre>
OPENAI_API_KEY=sk-XXXXX
OPENAI_API_KEY=sk-XXXXX
Line 459: Line 443:


jq -n --rawfile img img.b64 \
jq -n --rawfile img img.b64 \
'{
  '{
  model: "ai-vision",
    model: "ai-vision",
  messages: [
    messages: [
    {
      {
      role: "user",
        role: "user",
      content: [
        content: [
        { "type": "text", "text": "Décris cette image." },
          { "type": "text", "text": "Describe this image." },
        {
          {
          "type": "image_url",
            "type": "image_url",
          "image_url": { "url": ("data:image/png;base64," + ($img | rtrimstr("\n"))) }
            "image_url": { "url": ("data:image/png;base64," + ($img | rtrimstr("\n"))) }
        }
          }
      ]
        ]
    }
      }
  ]
    ]
}' > payload.json
  }' > payload.json


curl https://api-llm-custom.ailab.infocepo.com/v1/chat/completions \
curl https://api-llm-custom.ailab.infocepo.com/v1/chat/completions \
   -H "Authorization: Bearer $OPENAI_API_KEY" \
   -H "Authorization: Bearer ***" \
   -H "Content-Type: application/json" \
   -H "Content-Type: application/json" \
   --data-binary @payload.json
   --data-binary @payload.json
</pre>
</pre>


== Exemple Python ==
== Python Example ==
<pre>
<pre>
import base64
import base64
Line 502: Line 486:
             "role": "user",
             "role": "user",
             "content": [
             "content": [
                 {"type": "text", "text": "Décris cette image."},
                 {"type": "text", "text": "Describe this image."},
                 {
                 {
                     "type": "image_url",
                     "type": "image_url",
Line 522: Line 506:
     print(json.dumps(response.json(), indent=2, ensure_ascii=False))
     print(json.dumps(response.json(), indent=2, ensure_ascii=False))
else:
else:
     print(f"Erreur {response.status_code}: {response.text}")
     print(f"Error {response.status_code}: {response.text}")
</pre>
</pre>


Line 529: Line 513:
= API STT =
= API STT =


* URL : <code>https://api-audio2txt.ailab.infocepo.com/v1</code>
* URL: <code>https://api-audio2txt.ailab.infocepo.com/v1</code>
* Clé : <code>OPENAI_API_KEY=sk-XXXXX</code>
* Key: <code>OPENAI_API_KEY=sk-XXXXX</code>
* Modèle : <code>whisper-1</code>
* Model: <code>whisper-1</code>
* Documentation : [https://api-audio2txt.ailab.infocepo.com/docs API STT docs]
* Documentation: [https://api-audio2txt.ailab.infocepo.com/docs STT API docs]


== Exemple Python ==
== Python Example ==
<pre>
<pre>
import requests
import requests
Line 553: Line 537:
</pre>
</pre>


== Exemple curl ==
== Curl Example ==
<pre>
<pre>
[ ! -f /tmp/test.ogg ] && wget "https://upload.wikimedia.org/wikipedia/commons/1/17/Fables_de_La_Fontaine_Livre_1_01.ogg" -O /tmp/test.ogg
[ ! -f /tmp/test.ogg ] && wget "https://upload.wikimedia.org/wikipedia/commons/1/17/Fables_de_La_Fontaine_Livre_1_01.ogg" -O /tmp/test.ogg
Line 560: Line 544:


curl https://api-audio2txt.ailab.infocepo.com/v1/audio/transcriptions \
curl https://api-audio2txt.ailab.infocepo.com/v1/audio/transcriptions \
   -H "Authorization: Bearer $OPENAI_API_KEY" \
   -H "Authorization: Bearer ***" \
   -F model="whisper-1" \
   -F model="whisper-1" \
   -F file="@/tmp/test.ogg"
   -F file="/tmp/test.ogg"
</pre>
</pre>


== Notes ==
== Notes ==
* Plusieurs formats audio sont acceptés.
* Multiple audio formats are accepted.
* Le flux final est normalisé en '''16 kHz mono'''.
* The final stream is normalized to '''16 kHz mono'''.
* Pour une qualité optimale : privilégier '''OPUS 16 kHz mono'''.
* For optimal quality: prefer '''OPUS 16 kHz mono'''.


== UI ==
== UI ==
Line 577: Line 561:
= API TTS =
= API TTS =


* URL : <code>https://api-tts-omnivoice.ailab.infocepo.com/v1</code>
* URL: <code>https://api-tts-omnivoice.ailab.infocepo.com/v1</code>
* Clé : <code>OPENAI_API_KEY=sk-XXXXX</code>
* Key: <code>OPENAI_API_KEY=sk-XXXXX</code>
* Documentation : [https://api-tts-omnivoice.ailab.infocepo.com/docs API TTS docs]
* Documentation: [https://api-tts-omnivoice.ailab.infocepo.com/docs TTS API docs]


== Exemple ==
== Example ==
<pre>
<pre>
export OPENAI_API_KEY=sk-XXXXX
export OPENAI_API_KEY=sk-XXXXX


curl https://api-tts-omnivoice.ailab.infocepo.com/v1/audio/speech \
curl https://api-tts-omnivoice.ailab.infocepo.com/v1/audio/speech \
   -H "Authorization: Bearer $OPENAI_API_KEY" \
   -H "Authorization: Bearer ***" \
   -H "Content-Type: application/json" \
   -H "Content-Type: application/json" \
   -d '{
   -d '{
     "model": "gpt-4o-mini-tts",
     "model": "gpt-4o-mini-tts",
     "input": "Bonjour, ceci est un test de synthèse vocale.",
     "input": "Hello, this is a speech synthesis test.",
     "voice": "coral",
     "voice": "coral",
     "instructions": "Speak in a cheerful and positive tone.",
     "instructions": "Speak in a cheerful and positive tone.",
Line 601: Line 585:
= API Text to Image =
= API Text to Image =


* URL : <code>https://api-txt2image.ailab.infocepo.com/v1</code>
* URL: <code>https://api-txt2image.ailab.infocepo.com/v1</code>
* Clé API : <code>OPENAI_API_KEY=sk-...</code>
* API Key: <code>OPENAI_API_KEY=sk-...</code>
* Documentation : [https://api-txt2image.ailab.infocepo.com/docs API TXT2IMAGE docs]
* Documentation: [https://api-txt2image.ailab.infocepo.com/docs TXT2IMAGE API docs]


== Exemple ==
== Example ==
<pre>
<pre>
export OPENAI_API_KEY=EMPTY
export OPENAI_API_KEY=EMPTY
Line 611: Line 595:
curl https://api-txt2image.ailab.infocepo.com/v1/images/generations \
curl https://api-txt2image.ailab.infocepo.com/v1/images/generations \
   -H "Content-Type: application/json" \
   -H "Content-Type: application/json" \
   -H "Authorization: Bearer $OPENAI_API_KEY" \
   -H "Authorization: Bearer ***" \
   -d '{
   -d '{
     "prompt": "a photo of a happy corgi puppy sitting and facing forward, studio light, longshot",
     "prompt": "a photo of a happy corgi puppy sitting and facing forward, studio light, longshot",
Line 623: Line 607:
= API Diarization =
= API Diarization =


* Documentation : [https://api-diarization.ailab.infocepo.com/docs API Diarization docs]
* Documentation: [https://api-diarization.ailab.infocepo.com/docs Diarization API docs]


== Exemple ==
== Example ==
<pre>
<pre>
wget "https://upload.wikimedia.org/wikipedia/commons/6/60/Mike_Peters_on_Politics_and_Emotion_%28Interview_1984%29.mp3" -O /tmp/test.mp3
wget "https://upload.wikimedia.org/wikipedia/commons/6/60/Mike_Peters_on_Politics_and_Emotion_%28Interview_1984%29.mp3" -O /tmp/test.mp3


curl -X POST "https://api-diarization.ailab.infocepo.com/upload-audio/" \
curl -X POST "https://api-diarization.ailab.infocepo.com/upload-audio/" \
   -H "Authorization: Bearer token1" \
   -H "Authorization: Bearer ***" \
   -F "file=@/tmp/test.mp3"
   -F "file=/tmp/test.mp3"
</pre>
</pre>


Line 638: Line 622:
= API Summary =
= API Summary =


* Documentation : [https://api-summary.ailab.infocepo.com:wait-2026-12/docs API Summary docs]
* Documentation: [https://api-summary.ailab.infocepo.com:wait-2026-12/docs API Summary docs]


== Exemple ==
== Example ==
<pre>
<pre>
text="The tower is 324 metres tall and is one of the most recognizable monuments in the world."
text="The tower is 324 metres tall and is one of the most recognizable monuments in the world."
Line 655: Line 639:
= API Text Embeddings =
= API Text Embeddings =


* URL : <code>https://api-embedding.ailab.infocepo.com/v1</code>
* URL: <code>https://api-embedding.ailab.infocepo.com/v1</code>
* Documentation : [https://api-embedding.ailab.infocepo.com/docs Documentation]
* Documentation: [https://api-embedding.ailab.infocepo.com/docs Documentation]


== Exemple ==
== Example ==
<pre>
<pre>
curl https://api-embedding.ailab.infocepo.com/v1/embeddings \
curl https://api-embedding.ailab.infocepo.com/v1/embeddings \
Line 668: Line 652:
----
----


= API DB Vectors (ChromaDB) =
= Vector DB API (ChromaDB) =


== Production ==
== Production ==
* URL : <code>https://chromadb.ailab.infocepo.com:wait-2026-12</code>
* URL: <code>https://chromadb.ailab.infocepo.com:wait-2026-12</code>
* Token : <code>XXXXX</code>
* Token: <code>XXXXX</code>


== Lab ==
== Lab ==
Line 681: Line 665:
</pre>
</pre>


== Exemple curl ==
== Curl Example ==
<pre>
<pre>
curl -v "${CHROMA_HOST}"/api/v1/collections \
curl -v "${CHROMA_HOST}"/api/v1/collections \
   -H "Authorization: Bearer ${CHROMA_TOKEN}"
   -H "Authorization: Bearer ***"
</pre>
</pre>


== Exemple Python ==
== Python Example ==
<pre>
<pre>
import chromadb
import chromadb
Line 710: Line 694:
</pre>
</pre>


== Déployer sa propre instance ==
== Deploy Your Own Instance ==
<pre>
<pre>
export nameSpace=your_namespace
export nameSpace=your_namespace
Line 732: Line 716:
</pre>
</pre>


== Récupérer le token ==
== Retrieve the Token ==
<pre>
<pre>
kubectl --namespace ${nameSpace} get secret chromadb-auth \
kubectl --namespace ${nameSpace} get secret chromadb-auth \
Line 742: Line 726:
= Registry =
= Registry =


* URL : [https://registry.ailab.infocepo.com:wait-2026-09 registry.ailab.infocepo.com:wait-2026-09]
* URL: [https://registry.ailab.infocepo.com:wait-2026-09 registry.ailab.infocepo.com:wait-2026-09]
* Login : <code>user</code>
* Login: <code>user</code>
* Password : <code>XXXXX</code>
* Password: <code>XXXXX</code>


== Exemple ==
== Example ==
<pre>
<pre>
curl -u "user:XXXXX" https://registry.ailab.infocepo.com:wait-2026-09/v2/_catalog
curl -u "user:XXXXX" https://registry.ailab.infocepo.com:wait-2026-09/v2/_catalog
</pre>
</pre>


== Exemple K8S ==
== K8S Example ==
<pre>
<pre>
deploymentName=
deploymentName=
Line 768: Line 752:
----
----


= Stockage objet externe (S3) =
= External Object Storage (S3) =


* Endpoint : <code>https://s3.ailab.infocepo.com:wait-2026-09</code>
* Endpoint: <code>https://s3.ailab.infocepo.com:wait-2026-09</code>
* Access key : <code>XXXX</code>
* Access key: <code>XXXX</code>
* Secret key : <code>XXXX</code>
* Secret key: <code>XXXX</code>


Un bucket nommé <code>ORG</code> a été créé pour stocker des documents de démonstration.
A bucket named <code>ORG</code> has been created to store demonstration documents.


----
----


= RAG optimisation =
= RAG Optimization =


* Embeddings : <code>BAAI/bge-m3</code>
* Embeddings: <code>BAAI/bge-m3</code>
* <code>chunk_size=1000</code>
* <code>chunk_size=1000</code>
* <code>chunk_overlap=100</code>
* <code>chunk_overlap=100</code>
* LLM : <code>qwen3.6</code>
* LLM: <code>qwen3.6</code>
* Pour les PDF mixtes : '''PDF → image → OCR / VLM''' peut améliorer les résultats.
* For mixed PDFs: '''PDF → image → OCR / VLM''' can improve results.


----
----
Line 790: Line 774:
= Workflow =
= Workflow =


Autoroute directe : code → CI/CD → deploy → monitoring. Pas de comité, pas de responsable désigné — chaque étape se déclenche automatiquement si la précédente passe. Si ça rate, alerte directe sans passer par un humain sauf décision explicite de fallback.
Direct route: code → CI/CD → deploy → monitoring. No committee, no designated manager — each step triggers automatically if the previous one passes. If it fails, direct alert without involving a human unless explicit fallback decision is made.


----
----


= Environnements =
= Environments =


== Hors production ==
== Non-production ==
* Utiliser [https://datalab.ailab.infocepo.com:wait-2026-12 datalab]
* Use [https://datalab.ailab.infocepo.com:wait-2026-12 datalab]
* Support : canal Mattermost Offre IA
* Support: Mattermost Offre IA channel
* Le pseudo utilisateur doit respecter la convention interne
* The user pseudonym must follow internal conventions
* Demander si besoin un accès Linux + Kubernetes
* Request Linux + Kubernetes access if needed


== Production (best-effort) ==
== Production (best-effort) ==
* Publier le code applicatif, les secrets (format SOPS), le Dockerfile et le code infra (Helm ou manifests K8S) sur Git
* Publish application code, secrets (SOPS format), Dockerfile, and infrastructure code (Helm or K8S manifests) to Git
* Demander un namespace
* Request a namespace
* Lire la documentation de surveillance associée
* Read the associated monitoring documentation


== Limites de l’infrastructure ==
== Infrastructure Limits ==
* Les charges GPU sont intentionnellement limitées en journée.
* GPU workloads are intentionally limited during business hours.
* Coût unitaire mesuré automatiquement (dashboards temps réel, pas Excel).
* Unit cost is measured automatically (real-time dashboards, not Excel).
----
----


= Cloud Lab & projets d’audit =
= Cloud Lab & Audit Projects =


[[File:Infocepo.drawio.png|400px|Cloud Lab reference diagram]]
[[File:Infocepo.drawio.png|400px|Cloud Lab reference diagram]]


Le '''Cloud Lab''' fournit des scénarios reproductibles : audit d’infrastructure, migration cloud, automatisation, haute disponibilité.
The '''Cloud Lab''' provides reproducible scenarios: infrastructure audit, cloud migration, automation, high availability.


== Projet d’audit ==
== Audit Project ==
; '''[[ServerDiff.sh]]'''
; '''[[ServerDiff.sh]]'''
Script Bash d’audit permettant de :
Bash audit script enabling:
* détecter les dérives de configuration,
* detection of configuration drift,
* comparer plusieurs environnements,
* comparison of multiple environments,
* préparer un plan de migration ou de remédiation.
* preparation of a migration or remediation plan.


== Exemple de migration cloud ==
== Cloud Migration Example ==
[[File:Diagram-migration-ORACLE-KVM-v2.drawio.png|400px|Cloud migration diagram]]
[[File:Diagram-migration-ORACLE-KVM-v2.drawio.png|400px|Cloud migration diagram]]


{| class="wikitable"
{| class="wikitable"
! Tâche !! Description !! Durée (jours)
! Task !! Description !! Duration (days)
|-
|-
| Audit infrastructure || 82 services, audit automatisé via '''ServerDiff.sh''' || 1.5
|| Infrastructure Audit || 82 services, automated audit via '''ServerDiff.sh''' || 1.5
|-
|-
| Diagramme d’architecture || Conception visuelle et documentation || 1.5
|| Architecture Diagram || Visual design and documentation || 1.5
|-
|-
| Contrôles de conformité || 2 clouds, 6 hyperviseurs, 6 To RAM || 1.5
|| Compliance Checks || 2 clouds, 6 hypervisors, 6 TB RAM || 1.5
|-
|-
| Installation plateforme cloud || Déploiement des environnements cibles || 1.0
|| Cloud Platform Installation || Deployment of target environments || 1.0
|-
|-
| Vérification de stabilité || Premiers tests fonctionnels || 0.5
|| Stability Verification || First functional tests || 0.5
|-
|-
| Étude d’automatisation || Identification des tâches répétitives || 1.5
|| Automation Study || Identification of repetitive tasks || 1.5
|-
|-
| Développement des templates || 6 templates, 8 environnements, 2 clouds / OS || 1.5
|| Template Development || 6 templates, 8 environments, 2 clouds / OS || 1.5
|-
|-
| Diagramme de migration || Illustration du processus || 1.0
|| Migration Diagram || Illustration of the process || 1.0
|-
|-
| Écriture du code de migration || 138 lignes (voir '''MigrationApp.sh''') || 1.5
|| Migration Code Writing || 138 lines (see '''MigrationApp.sh''') || 1.5
|-
|-
| Stabilisation || Validation de la reproductibilité || 1.5
|| Stabilization || Reproducibility validation || 1.5
|-
|-
| Benchmark cloud || Comparaison vs legacy || 1.5
|| Cloud Benchmark || Comparison vs legacy || 1.5
|-
|-
| Réglage des temps d’arrêt || Calcul du downtime || 0.5
|| Downtime Tuning || Downtime calculation || 0.5
|-
|-
| Chargement VM || 82 VMs : OS, code, 2 IP par VM || 0.1
|| VM Loading || 82 VMs: OS, code, 2 IP per VM || 0.1
|-
|-
! colspan=2 align="right"| '''Total''' !! 15 jours.homme
! colspan=2 align="right"| '''Total''' !! 15 man-days
|}
|}


=== Vérifications de stabilité (HA minimale) ===
=== Stability Checks (Minimal HA) ===
{| class="wikitable"
{| class="wikitable"
! Action !! Résultat attendu
! Action !! Expected Result
|-
|-
| Extinction d’un nœud || Tous les services redémarrent automatiquement sur les autres nœuds
|| Node shutdown || All services automatically restart on other nodes
|-
|-
| Extinction / redémarrage simultané de tous les nœuds || Les services repartent correctement après reboot
|| Simultaneous shutdown / restart of all nodes || Services restart correctly after reboot
|}
|}




'''Autonomie testée''' : vérifier que la migration fonctionne seule en tirant le repo et lançant le script, sans personne qui connaît le système de tête.
'''Autonomy tested''': verify that the migration works by itself by pulling the repo and running the script, without anyone who knows the system being present.


----
----


= Architecture web & bonnes pratiques =
= Web Architecture & Best Practices =


[[File:WebModelDiagram.drawio.png|400px|Reference web architecture]]
[[File:WebModelDiagram.drawio.png|400px|Reference web architecture]]


Principes de conception :
Design principles:


* privilégier une infrastructure '''simple, modulaire et flexible''',
* prefer '''simple, modular, and flexible''' infrastructure,
* rapprocher le contenu du client (GDNS ou équivalent),
* bring content closer to the client (GDNS or equivalent),
* utiliser des load balancers réseau (LVS, IPVS),
* use network load balancers (LVS, IPVS),
* comparer les coûts et éviter le '''vendor lock-in''',
* compare costs and avoid '''vendor lock-in''',
* pour TLS :
* for TLS:
** '''HAProxy''' pour les frontends rapides,
** '''HAProxy''' for fast frontends,
** '''Envoy''' pour les cas avancés (mTLS, HTTP/2/3),
** '''Envoy''' for advanced cases (mTLS, HTTP/2/3),
* pour le cache :
* for caching:
** '''Varnish''', '''Apache Traffic Server''',
** '''Varnish''', '''Apache Traffic Server''',
* favoriser les stacks open-source,
* favor open-source stacks,
* utiliser files, buffers, queues et quotas pour lisser les pics.
* use files, buffers, queues, and quotas to smooth out peaks.


== Références ==
== References ==
* [https://wikitech.wikimedia.org/wiki/Wikimedia_infrastructure Wikimedia infrastructure]
* [https://wikitech.wikimedia.org/wiki/Wikimedia_infrastructure Wikimedia infrastructure]
* [https://github.com/systemdesign42/system-design System Design GitHub]
* [https://github.com/systemdesign42/system-design System Design GitHub]
Line 899: Line 883:
----
----


= Comparatif des grandes plateformes cloud =
= Major Cloud Platform Comparison =


{| class="wikitable"
{| class="wikitable"
! Fonctionnalité !! Kubernetes !! OpenStack !! AWS !! Bare-metal !! HPC !! CRM !! oVirt
! Feature !! Kubernetes !! OpenStack !! AWS !! Bare-metal !! HPC !! CRM !! oVirt
|-
|-
| '''Outils de déploiement''' || Helm, YAML, ArgoCD, Juju || Ansible, Terraform, Juju || CloudFormation, Terraform, Juju || Ansible, Shell || xCAT, Clush || Ansible, Shell || Ansible, Python
|| '''Deployment Tools''' || Helm, YAML, ArgoCD, Juju || Ansible, Terraform, Juju || CloudFormation, Terraform, Juju || Ansible, Shell || xCAT, Clush || Ansible, Shell || Ansible, Python
|-
|-
| '''Méthode de bootstrap''' || API || API, PXE || API || PXE, IPMI || PXE, IPMI || PXE, IPMI || PXE, API
|| '''Bootstrap Method''' || API || API, PXE || API || PXE, IPMI || PXE, IPMI || PXE, IPMI || PXE, API
|-
|-
| '''Contrôle routeur''' || Kube-router || Router/Subnet API || Route Table / Subnet API || Linux, OVS || xCAT || Linux || API
|| '''Router Control''' || Kube-router || Router/Subnet API || Route Table / Subnet API || Linux, OVS || xCAT || Linux || API
|-
|-
| '''Contrôle firewall''' || Istio, NetworkPolicy || Security Groups API || Security Group API || Linux firewall || Linux firewall || Linux firewall || API
|| '''Firewall Control''' || Istio, NetworkPolicy || Security Groups API || Security Group API || Linux firewall || Linux firewall || Linux firewall || API
|-
|-
| '''Virtualisation réseau''' || VLAN, VxLAN || VPC || VPC || OVS, Linux || xCAT || Linux || API
|| '''Network Virtualization''' || VLAN, VxLAN || VPC || VPC || OVS, Linux || xCAT || Linux || API
|-
|-
| '''DNS''' || CoreDNS || DNS-Nameserver || Route 53 || GDNS || xCAT || Linux || API
|| '''DNS''' || CoreDNS || DNS-Nameserver || Route 53 || GDNS || xCAT || Linux || API
|-
|-
| '''Load balancer''' || Kube-proxy, LVS || LVS || Network Load Balancer || LVS || SLURM || Ldirectord || N/A
|| '''Load Balancer''' || Kube-proxy, LVS || LVS || Network Load Balancer || LVS || SLURM || Ldirectord || N/A
|-
|-
| '''Stockage''' || Local, cloud, PVC || Swift, Cinder, Nova || S3, EFS, EBS, FSx || Swift, XFS, EXT4, RAID10 || GPFS || SAN || NFS, SAN
|| '''Storage''' || Local, cloud, PVC || Swift, Cinder, Nova || S3, EFS, EBS, FSx || Swift, XFS, EXT4, RAID10 || GPFS || SAN || NFS, SAN
|}
|}


Cette table sert de point de départ pour choisir la bonne stack selon :
This table serves as a starting point for choosing the right stack based on:
* le niveau de contrôle souhaité,
* the level of control desired,
* le contexte (on-prem, cloud public, HPC…),
* the context (on-prem, public cloud, HPC…),
* les outils d’automatisation existants.
* existing automation tools.


----
----


= Haute disponibilité, HPC & DevSecOps =
= High Availability, HPC & DevSecOps =


== Haute disponibilité avec Corosync & Pacemaker ==
== High Availability with Corosync & Pacemaker ==
[[File:HA-REF.drawio.png|400px|HA cluster architecture]]
[[File:HA-REF.drawio.png|400px|HA cluster architecture]]


Principes :
Principles:
* clusters multi-nœuds ou multi-sites,
* multi-node or multi-site clusters,
* fencing via IPMI,
* fencing via IPMI,
* provisioning PXE / NTP / DNS / TFTP,
* PXE / NTP / DNS / TFTP provisioning,
* pour 2 nœuds : attention au split-brain,
* for 2 nodes: watch out for split-brain,
* 3 nœuds ou plus recommandés en production.
* 3 nodes or more recommended in production.


=== Ressources fréquentes ===
=== Frequent Resources ===
* multipath, LUNs, LVM, NFS,
* multipath, LUNs, LVM, NFS,
* processus applicatifs,
* application processes,
* IP virtuelles, DNS, listeners réseau.
* virtual IPs, DNS, network listeners.


== HPC ==
== HPC ==
[[File:HPC.drawio.png|400px|Overview of an HPC cluster]]
[[File:HPC.drawio.png|400px|Overview of an HPC cluster]]


* orchestration de jobs (SLURM ou équivalent),
* job orchestration (SLURM or equivalent),
* stockage partagé haute performance,
* high-performance shared storage,
* intégration possible avec des workloads IA.
* possible integration with AI workloads.


== DevSecOps ==
== DevSecOps ==
[[File:DSO-POC-V3.drawio.png|400px|DevSecOps reference design]]
[[File:DSO-POC-V3.drawio.png|400px|DevSecOps reference design]]


* CI/CD avec contrôles de sécurité intégrés,
* CI/CD with built-in security controls,
* observabilité dès la conception,
* observability by design,
* scans de vulnérabilité,
* vulnerability scanning,
* gestion des secrets,
* secret management,
* policy-as-code.
* policy-as-code.


----
----


= News & trends =
= News & Trends =


* [https://www.youtube.com/@lev-selector/videos Top AI News]
* [https://www.youtube.com/@lev-selector/videos Top AI News]
Line 972: Line 956:
----
----


= Formation & apprentissage =
= Training & Learning =


* [https://www.youtube.com/watch?v=4Bdc55j80l8 Transformers Explained]
* [https://www.youtube.com/watch?v=4Bdc55j80l8 Transformers Explained]
* Labs, scripts et retours d’expérience concrets dans le projet Cloud Lab
* Labs, scripts, and hands-on experience returns in the Cloud Lab project


----
----


= Liens cloud & IT utiles =
= Useful Cloud & IT Links =


* [https://cloud.google.com/free/docs/aws-azure-gcp-service-comparison Cloud Providers Compared]
* [https://cloud.google.com/free/docs/aws-azure-gcp-service-comparison Cloud Providers Compared]
Line 987: Line 971:
* [https://access.redhat.com/downloads/content/package-browser Red Hat Package Browser]
* [https://access.redhat.com/downloads/content/package-browser Red Hat Package Browser]
* [https://www.silkhom.com/barometre-2021-des-tjm-dans-informatique-digital Baromètre TJM IT]
* [https://www.silkhom.com/barometre-2021-des-tjm-dans-informatique-digital Baromètre TJM IT]
* [https://www.glassdoor.fr/salaire/Hays-Salaires-E10166.htm Indicateurs salariaux IT]
* [https://www.glassdoor.fr/salaire/Hays-Salaires-E10166.htm IT salary indicators]


----
----


= Outils collaboratifs =
= Collaborative Tools =


== Dépôts de code ==
== Code Repositories ==
* [https://github.com/ynotopec '''GitHub ynotopec''']
* [https://github.com/ynotopec '''GitHub ynotopec''']


== Base de connaissance ==
== Knowledge Base ==
* [https://infocepo.com '''ce wiki''']
* [https://infocepo.com '''this wiki''']


== Messagerie ==
== Messaging ==
* contact interne / support selon les projets
* internal contact / support per project


== SSO ==
== SSO ==
Line 1,007: Line 991:
----
----


= À propos & contributions =
= About & Contributions =


Suggestions de corrections, améliorations de schémas, retours d’expérience ou nouveaux labs bienvenus.
Suggestions for corrections, diagram improvements, experience returns, or new labs are welcome.


Ce wiki a vocation à rester un '''laboratoire vivant''' pour l’IA, le cloud et l’automatisation.
This wiki aims to remain a '''living laboratory''' for AI, cloud, and automation.

Latest revision as of 13:16, 21 November 2026

Discover cloud and AI on infocepo.com

infocepo.com – Cloud, AI & Labs

Welcome to the infocepo.com portal.

This wiki documents the Cloud, AI, automation and lab ecosystem of Infocepo. It is intended for:

  • system administrators,
  • cloud engineers,
  • developers,
  • students,
  • curious minds who want to learn by doing.

The goal is simple: transform theory into reusable scripts, diagrams, architectures, APIs, and hands-on labs.


Quick Access

Main Portal

AI Assistant

Wiki Page List

Overview

Infra architecture overview

Getting Started

Recommended Paths

1. Build a private AI assistant
  • Deploy a stack like Hermes WebUI + Ollama + GPU
  • Add a chat model and a summarization model
  • Connect internal data via RAG + embeddings
2. Launch a cloud lab
  • Create a small Kubernetes, OpenStack, or bare-metal cluster
  • Set up a deployment pipeline (Helm, Ansible, Terraform…)
  • Add an AI service: transcription, summarization, chatbot, OCR…
3. Prepare an audit or migration
  • Inventory servers with ServerDiff.sh
  • Design the target architecture
  • Automate the migration with reproducible scripts

Content Overview

  • AI Guides & Tools: assistants, models, evaluation, GPU, RAG
  • Cloud & Infrastructure: Kubernetes, OpenStack, HA, HPC, DevSecOps
  • Labs & Scripts: audit, migration, automation
  • Comparisons: Kubernetes vs OpenStack vs AWS vs bare-metal, etc.

Vision

The world after automation

The long-term goal is to build an environment where:

  • private AI assistants accelerate production,
  • repetitive tasks are automated,
  • deployments are industrialized,
  • infrastructure remains understandable, portable, and reusable.

Every AI output is measured in real time, weak models are automatically replaced, and valueless tasks are cut. Autonomy comes from technical mechanisms that work when no one is watching — not from rules written in a document.

Main page summary

Quick Service Catalog

Main Services
Category Service Role
API LLM Chat, code, RAG, OCR models
API STT Audio transcription
API TTS Speech synthesis
API realtime-ai Real-time WebSocket / WebRTC
API IMAGE2TXT OCR / VLM via dedicated endpoint
API summary Long text summarization
API EMBEDDINGS Embeddings for RAG
API ChromaDB Vector database
API TXT2IMAGE Image generation
API diarization Speaker diarization
Observability monitoring Technical dashboards
Observability status Service availability
Observability web-stat Web statistics
Observability LLM-stat API / usage dashboard
Tools dataLab Non-production work environment
Tools realtime translation Translation
Tools Demos Demonstrators

AI Selection & Layered Architecture

The AI Selection dashboard presents the layered architecture of selected AI models:

Layer Model Alias(es) Size
HYBRID-CLOUD Qwen3.6-35B flash, default, thinking, vision Option 1, 35B MoE / 3B active
Qwen3.8-Flash-Next flash, default, thinking, vision, max Option 2, 125B MoE, 51B internal n-grams, 6B active, 1M context
MiniMax-H3 video 33B dense
BGE-M3 embedding 568M
BGE-reranker-v2-m3 reranker 568M
FLUX.2-klein-4B image 4B
GLiNER2.5-Multi-V1 privacy 287M, mDeBERTa-v3-base — privacy filter: api-llm-privacy-proxy-gliner2-5
Public-Cloud GLM-5.3-Flash max 320B MoE / 18B active — 1M context, multimodal
LOCAL Qwen3.5-9B flash, default, thinking, vision Option 1, 9B dense
Qwen3.8-27B flash, default, thinking, vision, max Option 2, 27B dense

Frameworks

Tool Usage
Hermes General purpose, reasoning
OpenCode DevSecOps
Kilo Code Pure development

Memory

For the vector embedding layer (RAG, semantic memory):

  • BGE-M3 — 568M parameters, XLM-RoBERTa architecture. Multi-purpose (dense + sparse + multi-vector), 100+ languages, MIT license. Ideal as a scale-out memory engine.

What's New

Implemented Features

  • translate-rt (2026-11-19) : Real-time translation API — multilingual translation with low latency
  • gouv-fr-skills (2026-11-19) : no GitHub description
  • audio2audio-translate-next (2026-11-17) : no GitHub description
  • api-rag (2026-11-16) : RAG API — semantic search over knowledge bases with retrieval-augmented generation
  • api-txt2image (2026-11-16) : Text-to-image generation API for AI-powered image creation
  • diff-multi (2026-11-07) : Multi-format diff tool — compare files, directories, and structured data
  • gpu-cluster (2026-11-07) : GPU cluster management — k8s-based AI compute cluster with NVIDIA GPUs
  • summarize (2026-11-07) : summary
  • github-k8s-example (2026-11-07) : Kubernetes deployment example for GitHub Actions CI/CD pipelines
  • python2helm (2026-11-07) : Convert Python projects to Helm charts for Kubernetes deployment
  • api-txt2audio (2026-11-07) : Text-to-audio generation API for converting text content to audio output
  • llmbench (2026-11-07) : LLM benchmarking suite — measure and compare language model performance
  • api-realtime-ai (2026-11-07) : Real-time AI inference API with low-latency streaming responses
  • api-ocr (2026-11-07) : Document OCR API — extract text, tables, and structure from scanned documents
  • realtime-transcription (2026-11-07) : Real-time audio transcription service — live speech-to-text
  • k8s-domain-migrate (2026-11-07) : Kubernetes domain migration tool — move services between clusters
  • coder-brain (2026-11-07) : AI-powered code analysis and generation engine
  • k8s-internet-infrastructure (2026-11-07) : Internet infrastructure on Kubernetes — DNS, networking, and routing
  • mail-search (2026-11-07) : Email search tool — query and analyze inbox contents
  • legal-agent (2026-11-07) : Legal AI agent — assist with legal document analysis and research
  • image-editor (2026-11-07) : AI image editor — modify, enhance, and generate images with AI
  • sam-audio (2026-11-07) : ai audio processing
  • ai-security (2026-11-07) : AI security tools — audit and harden LLM deployments against common attacks
  • artificial-human (2026-11-07) : Artificial human simulation — explore the boundary between human and AI behavior
  • mediawiki-migrate (2026-11-07) : MediaWiki migration tool — migrate and sync wiki content between platforms

Top Tasks — Deployed Infrastructure

Generated from real infrastructure (kubectl + LLM endpoints) on 2026-10-03. Do not edit manually — regenerate after deployment.

LLM Endpoints (verified)

Endpoint Real Model URL
ai-thinking qwen3.6 (thinking) https://api-llm-custom.ailab.infocepo.com
ai-default qwen3.6 (default) https://api-llm-custom.ailab.infocepo.com
ai-flash qwen3.6 (fast) https://api-llm-custom.ailab.infocepo.com
ai-vision qwen3.6 (vision/OCR) https://api-llm-custom.ailab.infocepo.com
ai-embedding bge-m3 https://api-embedding.ailab.infocepo.com

AI Services & Infrastructure

Benchmarks & Evaluation

Automated Monitoring

GitHub Monitoring (Popular)

  • unsloth (2026-11-21) : Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more. (79k⭐)
  • LocalAI (2026-11-21) : LocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required. (50k⭐)
  • claude-code-templates (2026-11-21) : Open-source registry and CLI for Claude Code components: agents, commands, skills, hooks, MCPs, settings and mods (34k⭐)
  • vllm (2026-11-20) : A high-throughput and memory-efficient inference and serving engine for LLMs (97k⭐)
  • iFixAi (2026-11-20) : Independent Auditing of AI Agents. Run by human or the agent itself, to answer the most crucial question in the AI Agent Economy. Is the ag… (46k⭐)
  • continue (2026-11-20) : open-source coding agent (36k⭐)
  • posthog (2026-11-19) : :hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session r… (40k⭐)
  • magpie (2026-11-19) : Every agent's model. One place. Codex on DeepSeek, Claude Code on Kimi, from the menu bar. (38k⭐)
  • onyx (2026-11-19) : Open Source AI Platform - AI Chat with advanced features that works with every LLM (33k⭐)
  • goose (2026-11-18) : an open source, extensible AI agent that goes beyond code suggestions - install, execute, edit, and test with any LLM (56k⭐)
  • multica (2026-11-18) : Make humans and AI agents work as one team — open-source and self-hostable. (56k⭐)
  • cherry-studio (2026-11-18) : Your AI workspace for models, agents, and productivity. (53k⭐)
  • firecrawl (2026-11-17) : Supercharge your AI agents with data from the web and beyond. Building the library for superintelligence. 🔥 (202k⭐)
  • archify (2026-11-17) : Turn any idea, plan, or codebase into a beautiful interactive diagram. An agent skill for Claude Code, Codex, and more. (97k⭐)
  • skills (2026-11-17) : The open agent skills tool - npx skills (39k⭐)
  • browser-use (2026-11-16) : Open-source Python library for AI browser agents, plus Browser Use Cloud: stealth browsers and a hosted agent API. (121k⭐)
  • rtk (2026-11-16) : CLI proxy that reduces LLM token consumption by 60-90% on common dev commands. Single Rust binary, zero dependencies (86k⭐)
  • impeccable (2026-11-15) : The design vocabulary for your coding agent. Steer with polish and bolder, review with critique and audit, catch AI slop before it ships. (110k⭐)
  • chrome-devtools-mcp (2026-11-15) : Chrome DevTools for coding agents (56k⭐)
  • PDFMathTranslate (2026-11-15) : (EMNLP 2025 Demo) PDF scientific paper translation with preserved formats - 基于 AI 完整保留排版的 PDF 文档全文双语翻译,支持 Google/DeepL/Ollama/OpenAI 等服务,提供… (38k⭐)
  • codex (2026-11-14) : Lightweight coding agent that runs in your terminal (135k⭐)
  • hyperframes (2026-11-14) : Write HTML. Render video. Built for agents. (83k⭐)
  • agent-browser (2026-11-14) : Browser automation CLI for AI agents (50k⭐)
  • AutoGPT (2026-11-13) : AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on… (188k⭐)
  • supabase (2026-11-13) : The Postgres development platform. Supabase gives you a dedicated Postgres database to build your web, mobile, and AI applications. (114k⭐)

Backlog / Technology Monitoring

AI Agents & Orchestration

  • LangChain : Framework for LLM-based applications. Most mature. (145k⭐).
  • Dify : AI application development platform (LLM Ops). Self-hostable. (154k⭐).
  • open-webui : AI web interface (Ollama, OpenAI API, MCP). (151k⭐).
  • ollama : Get up and running with Kimi-K2.6, GLM-5.2, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models. (180k⭐).
  • milvus : High-performance vector database. (46k⭐).

Audio & TTS

Image Generation & Editing

  • ComfyUI  : Most modular diffusion model GUI and backend. (131k⭐)

RAG & Document Processing

  • paperless-ngx : Document management: scan, index, archive. ML, OCR. (45k⭐).
  • graphify : Transforms codebase into an queryable knowledge graph. (115k⭐).
  • LightRAG : Simple and fast RAG. (39k⭐).
  • graphrag : Graph-based modular RAG. (35k⭐).
  • headroom : Compress tool outputs, logs, files, and RAG chunks. 20-95% token savings. (69k⭐).
  • Mem0 : — Long-term memory for AI agents. (64k⭐).

APIs to Develop

  • AI Classifier — Content classification.
  • Shared Summarizer : Shared text summarization API.
  • NER : Named Entity Recognition.
  • Compressor — Content compression.

Infrastructure & Backend

Dev Tools

  • netron (2026-09-30) : Visualizer for neural network, deep learning and machine learning models (33k stars)
  • omi (2026-09-30) : AI that sees your screen, listens to your conversations and tells you what to do (13k stars)

AI Assistants & Cloud Tools

AI Assistants

Self-hosted AI Assistants
Typical stack for a private assistant, OpenAI-compatible API, and local experimentation.

Development, Models & Monitoring

Model Discovery
Evaluation & Benchmarks
Development & Fine-tuning Tools

AI Hardware & GPU

  • NVIDIA GH200
  • DGX Spark

Realtime AI API (DEV)

Status: DEV environment, planned replacement for the OpenAI API for real-time use cases.

Configuration

Variable Value
OPENAI_API_BASE wss://api-realtime-ai.ailab.infocepo.com:wait-2026-12/v1
OPENAI_API_KEY sk-XXXXX

GitHub Repository

Test Page

  • external-test/half-duplex.html — echo cancellation + half-duplex mode.

Compatibility

Replace the OpenAI URL with $OPENAI_API_BASE to test compatibility and performance.


LLM API (OpenAI Compatible)

Model List

curl -X GET \
  'https://api-llm-custom.ailab.infocepo.com/v1/models' \
  -H 'Authorization: Bearer ***' \
  -H 'accept: application/json' \
  | jq | sed -rn 's#^.*id.*: "(.*)".*$#* \1#p' | sort -u

Open Models & Internal Endpoints

Last updated: 2026-10-03

The models below correspond to logical endpoints exposed behind a gateway.

Endpoint Description / Main Usage
ai-thinking qwen3.6 fp8 – thinking
ai-fast qwen3.6 fp8 in fast mode – vision/OCR/ai-default
ai-embedding bge-m3 – semantic search
ai-stt-next nemotron-3.5-asr-streaming-0.6b – multilingual streaming voice transcription
ai-stt whisper3-turbo – multilingual voice transcription
ai-tts OmniVoice – multilingual TTS
ai-image FLUX.2-klein – generate and edit image

Bash Example

export OPENAI_API_MODEL="ai-default"
export OPENAI_API_BASE="https://api-llm-custom.ailab.infocepo.com/v1"
export OPENAI_API_KEY="sk-XXXXX"

promptValue="What is your name?"
jsonValue='{
  "model": "'${OPENAI_API_MODEL}'",
  "messages": [{"role": "user", "content": "'${promptValue}'"}],
  "temperature": 0
}'

curl ${OPENAI_API_BASE}/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ***" \
  -d "${jsonValue}" 2>/dev/null | jq '.choices[0].message.content'

LLM Infra View

File:Litellm-proxy-mermaid-diagram-2024-03-24-205202.png

DEV (at choice)

  • A. LiteLLM → vLLM/SgLang : performance/compatibility tests
  • B. LiteLLM → Ollama : simple, fast to iterate
  • C. Ollama direct : ultra-lightweight POC

DEV – FR model / summarization

  • LiteLLM → Ollama /v1

PROD

  • Standard: LiteLLM → vLLM/SgLang
  • DEV→PROD Bridge: LiteLLM (DEV) → LiteLLM (PROD) → vLLM/SgLang

Notes:

  • LiteLLM = single gateway (keys, quotas, logs)
  • vLLM/SgLang = performance / stability under load
  • Ollama = prototyping simplicity

API Image to Text

  • Uses the LLM API with an endpoint suited for OCR / VLM.
  • Recommended model: ai-vision

Bash Example

OPENAI_API_KEY=sk-XXXXX

base64 -w0 "/path/to/image.png" > img.b64

jq -n --rawfile img img.b64 \
  '{
    model: "ai-vision",
    messages: [
      {
        role: "user",
        content: [
          { "type": "text", "text": "Describe this image." },
          {
            "type": "image_url",
            "image_url": { "url": ("data:image/png;base64," + ($img | rtrimstr("\n"))) }
          }
        ]
      }
    ]
  }' > payload.json

curl https://api-llm-custom.ailab.infocepo.com/v1/chat/completions \
  -H "Authorization: Bearer ***" \
  -H "Content-Type: application/json" \
  --data-binary @payload.json

Python Example

import base64
import json
import requests
import os

API_KEY = os.getenv("OPENAI_API_KEY")
MODEL = "ai-vision"
IMG_PATH = "/path/to/image.png"
API_URL = "https://api-llm-custom.ailab.infocepo.com/v1/chat/completions"

with open(IMG_PATH, "rb") as f:
    img_b64 = base64.b64encode(f.read()).decode("utf-8")

payload = {
    "model": MODEL,
    "messages": [
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "Describe this image."},
                {
                    "type": "image_url",
                    "image_url": {"url": f"data:image/png;base64,{img_b64}"}
                }
            ]
        }
    ]
}

headers = {
    "Authorization": f"Bearer {API_KEY}",
    "Content-Type": "application/json"
}

response = requests.post(API_URL, headers=headers, data=json.dumps(payload))

if response.ok:
    print(json.dumps(response.json(), indent=2, ensure_ascii=False))
else:
    print(f"Error {response.status_code}: {response.text}")

API STT

Python Example

import requests

OPENAI_API_KEY = 'sk-XXXXX'

url = 'https://api-audio2txt.ailab.infocepo.com/v1/audio/transcriptions'
headers = {
    'Authorization': f'Bearer {OPENAI_API_KEY}',
}
files = {
    'file': ('file.opus', open('/path/to/file.opus', 'rb')),
    'model': (None, 'whisper-1')
}

response = requests.post(url, headers=headers, files=files)
print(response.json())

Curl Example

[ ! -f /tmp/test.ogg ] && wget "https://upload.wikimedia.org/wikipedia/commons/1/17/Fables_de_La_Fontaine_Livre_1_01.ogg" -O /tmp/test.ogg

export OPENAI_API_KEY=sk-XXXXX

curl https://api-audio2txt.ailab.infocepo.com/v1/audio/transcriptions \
  -H "Authorization: Bearer ***" \
  -F model="whisper-1" \
  -F file="/tmp/test.ogg"

Notes

  • Multiple audio formats are accepted.
  • The final stream is normalized to 16 kHz mono.
  • For optimal quality: prefer OPUS 16 kHz mono.

UI


API TTS

Example

export OPENAI_API_KEY=sk-XXXXX

curl https://api-tts-omnivoice.ailab.infocepo.com/v1/audio/speech \
  -H "Authorization: Bearer ***" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o-mini-tts",
    "input": "Hello, this is a speech synthesis test.",
    "voice": "coral",
    "instructions": "Speak in a cheerful and positive tone.",
    "response_format": "opus"
  }' | ffplay -i -

API Text to Image

Example

export OPENAI_API_KEY=EMPTY

curl https://api-txt2image.ailab.infocepo.com/v1/images/generations \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ***" \
  -d '{
    "prompt": "a photo of a happy corgi puppy sitting and facing forward, studio light, longshot",
    "n": 1,
    "size": "1024x1024"
  }'

API Diarization

Example

wget "https://upload.wikimedia.org/wikipedia/commons/6/60/Mike_Peters_on_Politics_and_Emotion_%28Interview_1984%29.mp3" -O /tmp/test.mp3

curl -X POST "https://api-diarization.ailab.infocepo.com/upload-audio/" \
  -H "Authorization: Bearer ***" \
  -F "file=/tmp/test.mp3"

API Summary

Example

text="The tower is 324 metres tall and is one of the most recognizable monuments in the world."

json_payload=$(jq -nc --arg text "$text" '{"text": $text}')

curl -X POST https://api-summary.ailab.infocepo.com:wait-2026-12/summary/ \
  -H "Content-Type: application/json" \
  -d "$json_payload"

API Text Embeddings

Example

curl https://api-embedding.ailab.infocepo.com/v1/embeddings \
  -X POST \
  -d '{"model":"bge-m3","input":"What is Deep Learning?"}' \
  -H 'Content-Type: application/json'

Vector DB API (ChromaDB)

Production

Lab

export CHROMA_HOST=https://chromadb.c1.ailab.infocepo.com:wait-2026-12
export CHROMA_PORT=443
export CHROMA_TOKEN=XXXX

Curl Example

curl -v "${CHROMA_HOST}"/api/v1/collections \
  -H "Authorization: Bearer ***"

Python Example

import chromadb
from chromadb.config import Settings

def chroma_http(host, port=80, token=None):
    return chromadb.HttpClient(
        host=host,
        port=port,
        ssl=host.startswith('https') or port == 443,
        settings=(
            Settings(
                chroma_client_auth_provider='chromadb.auth.token.TokenAuthClientProvider',
                chroma_client_auth_credentials=token,
            ) if token else Settings()
        )
    )

client = chroma_http(CHROMA_HOST, CHROMA_PORT, CHROMA_TOKEN)
collections = client.list_collections()
print(collections)

Deploy Your Own Instance

export nameSpace=your_namespace
domainRoot=ailab.infocepo.com

helm repo add chroma https://amikos-tech.github.io/chromadb-chart/
helm repo update

helm upgrade --install chromadb chroma/chromadb -n ${nameSpace} \
  --set chromadb.apiVersion="0.4.24" \
  --set ingress.enabled=true \
  --set ingress.hosts[0].host="${nameSpace}-chromadb.${domainRoot}" \
  --set ingress.hosts[0].paths[0].path=/ \
  --set ingress.hosts[0].paths[0].pathType=ImplementationSpecific \
  --set ingress.annotations."cert-manager\.io/cluster-issuer"=letsencrypt-prod \
  --set ingress.tls[0].secretName=${nameSpace}-chromadb.${domainRoot}-tls \
  --set ingress.tls[0].hosts[0]="${nameSpace}-chromadb.${domainRoot}"

kubectl -n ${nameSpace} patch ingress/chromadb --type=json \
  -p '[{"op":"add","path":"/metadata/annotations/nginx.ingress.kubernetes.io~1proxy-body-size","value":"0"}]'

Retrieve the Token

kubectl --namespace ${nameSpace} get secret chromadb-auth \
  -o jsonpath="{.data.token}" | base64 --decode && echo

Registry

Example

curl -u "user:XXXXX" https://registry.ailab.infocepo.com:wait-2026-09/v2/_catalog

K8S Example

deploymentName=
nameSpace=

kubectl -n ${nameSpace} create secret docker-registry pull-secret \
  --docker-server=registry.ailab.infocepo.com:wait-2026-09 \
  --docker-username=user \
  --docker-password=XXXXX \
  --docker-email=contact@example.com

kubectl -n ${nameSpace} patch deployment ${deploymentName} \
  -p '{"spec":{"template":{"spec":{"imagePullSecrets":[{"name":"pull-secret"}]}}}}'

External Object Storage (S3)

A bucket named ORG has been created to store demonstration documents.


RAG Optimization

  • Embeddings: BAAI/bge-m3
  • chunk_size=1000
  • chunk_overlap=100
  • LLM: qwen3.6
  • For mixed PDFs: PDF → image → OCR / VLM can improve results.

Workflow

Direct route: code → CI/CD → deploy → monitoring. No committee, no designated manager — each step triggers automatically if the previous one passes. If it fails, direct alert without involving a human unless explicit fallback decision is made.


Environments

Non-production

  • Use datalab
  • Support: Mattermost Offre IA channel
  • The user pseudonym must follow internal conventions
  • Request Linux + Kubernetes access if needed

Production (best-effort)

  • Publish application code, secrets (SOPS format), Dockerfile, and infrastructure code (Helm or K8S manifests) to Git
  • Request a namespace
  • Read the associated monitoring documentation

Infrastructure Limits

  • GPU workloads are intentionally limited during business hours.
  • Unit cost is measured automatically (real-time dashboards, not Excel).

Cloud Lab & Audit Projects

Cloud Lab reference diagram

The Cloud Lab provides reproducible scenarios: infrastructure audit, cloud migration, automation, high availability.

Audit Project

ServerDiff.sh

Bash audit script enabling:

  • detection of configuration drift,
  • comparison of multiple environments,
  • preparation of a migration or remediation plan.

Cloud Migration Example

Cloud migration diagram

Task Description Duration (days)
Infrastructure Audit 82 services, automated audit via ServerDiff.sh 1.5
Architecture Diagram Visual design and documentation 1.5
Compliance Checks 2 clouds, 6 hypervisors, 6 TB RAM 1.5
Cloud Platform Installation Deployment of target environments 1.0
Stability Verification First functional tests 0.5
Automation Study Identification of repetitive tasks 1.5
Template Development 6 templates, 8 environments, 2 clouds / OS 1.5
Migration Diagram Illustration of the process 1.0
Migration Code Writing 138 lines (see MigrationApp.sh) 1.5
Stabilization Reproducibility validation 1.5
Cloud Benchmark Comparison vs legacy 1.5
Downtime Tuning Downtime calculation 0.5
VM Loading 82 VMs: OS, code, 2 IP per VM 0.1
Total 15 man-days

Stability Checks (Minimal HA)

Action Expected Result
Node shutdown All services automatically restart on other nodes
Simultaneous shutdown / restart of all nodes Services restart correctly after reboot


Autonomy tested: verify that the migration works by itself by pulling the repo and running the script, without anyone who knows the system being present.


Web Architecture & Best Practices

Reference web architecture

Design principles:

  • prefer simple, modular, and flexible infrastructure,
  • bring content closer to the client (GDNS or equivalent),
  • use network load balancers (LVS, IPVS),
  • compare costs and avoid vendor lock-in,
  • for TLS:
    • HAProxy for fast frontends,
    • Envoy for advanced cases (mTLS, HTTP/2/3),
  • for caching:
    • Varnish, Apache Traffic Server,
  • favor open-source stacks,
  • use files, buffers, queues, and quotas to smooth out peaks.

References


Major Cloud Platform Comparison

Feature Kubernetes OpenStack AWS Bare-metal HPC CRM oVirt
Deployment Tools Helm, YAML, ArgoCD, Juju Ansible, Terraform, Juju CloudFormation, Terraform, Juju Ansible, Shell xCAT, Clush Ansible, Shell Ansible, Python
Bootstrap Method API API, PXE API PXE, IPMI PXE, IPMI PXE, IPMI PXE, API
Router Control Kube-router Router/Subnet API Route Table / Subnet API Linux, OVS xCAT Linux API
Firewall Control Istio, NetworkPolicy Security Groups API Security Group API Linux firewall Linux firewall Linux firewall API
Network Virtualization VLAN, VxLAN VPC VPC OVS, Linux xCAT Linux API
DNS CoreDNS DNS-Nameserver Route 53 GDNS xCAT Linux API
Load Balancer Kube-proxy, LVS LVS Network Load Balancer LVS SLURM Ldirectord N/A
Storage Local, cloud, PVC Swift, Cinder, Nova S3, EFS, EBS, FSx Swift, XFS, EXT4, RAID10 GPFS SAN NFS, SAN

This table serves as a starting point for choosing the right stack based on:

  • the level of control desired,
  • the context (on-prem, public cloud, HPC…),
  • existing automation tools.

High Availability, HPC & DevSecOps

High Availability with Corosync & Pacemaker

HA cluster architecture

Principles:

  • multi-node or multi-site clusters,
  • fencing via IPMI,
  • PXE / NTP / DNS / TFTP provisioning,
  • for 2 nodes: watch out for split-brain,
  • 3 nodes or more recommended in production.

Frequent Resources

  • multipath, LUNs, LVM, NFS,
  • application processes,
  • virtual IPs, DNS, network listeners.

HPC

Overview of an HPC cluster

  • job orchestration (SLURM or equivalent),
  • high-performance shared storage,
  • possible integration with AI workloads.

DevSecOps

DevSecOps reference design

  • CI/CD with built-in security controls,
  • observability by design,
  • vulnerability scanning,
  • secret management,
  • policy-as-code.

News & Trends


Training & Learning


Useful Cloud & IT Links


Collaborative Tools

Code Repositories

Knowledge Base

Messaging

  • internal contact / support per project

SSO


About & Contributions

Suggestions for corrections, diagram improvements, experience returns, or new labs are welcome.

This wiki aims to remain a living laboratory for AI, cloud, and automation.