The AI community of Strasbourg

Lea Learn : EdTech Agents

Youri Durand (founder & CTO, Lea Learn), Fadi Sultan & Julien Brodier — Digital Village Strasbourg

Agentic RAG for matching and funding professional training: from a vaguely worded need to a funded course — qualification chatbot, scraping agents, embeddings (pgvector, HNSW), hybrid search (HyDE, reranking), OPCO rule base and SIRET → IDCC → OPCO resolution. Product, market and tech lessons from an EdTech startup.

RobotKraft — Hackathon IA & robotique (Hugging Face LeRobot)

Alsace Digitale × Strasbourg IA — FABEON, Illkirch-Graffenstaden

A 3-day hackathon (Oct 2–4): models trained by the teams drive SO-ARM100 robotic arms with the Hugging Face LeRobot ecosystem — vision, imitation learning, autonomy. Final challenge open to the public on Sunday.

Graph/RAG & NeSy AI : du doc brut au raisonnement

Julien Brodier — Talium

From raw documents to reasoning: a full pipeline covering extraction (OCR, VLMs), entity recognition (NER), knowledge graph construction (RDF vs Neo4j), GraphRAG querying strategies, and neuro-symbolic reasoning — combining the flexibility of LLMs with the rigor of symbolic AI.

Donner sa langue au « chat »

Mathieu Cordebard — web developer & WebRTC architect

A subjective, non-specialist take on how AIs work and what they change: how these models learn, what they cost (data, energy), and how they reshape the way we create, work, and think.

DUST : Deploying AI Agents Day-to-Day

Léandre Le Bizec — Solutions Engineer at Dust

Online session (English). How teams ship AI agents day-to-day across support, ops, finance, and engineering. Live demo: building an agent from scratch, plugged into existing tools and data, followed by Q&A.

AI For Chemistry — Build an LLM with agents for Chemistry

Teodora Djikic-Stojsic, Emil Sayilov & Anjila Ben Frej — Research Engineers, iCube

English session. Intro to medicinal chemistry and drug discovery, then a chemistry-specialised LLM: structured reasoning, analogue-series datasets, and an agentic layer with tool-calling to assist molecule research.

Créez des agents IA avec Google ADK

Sonny Mupfuni — AI engineer & PyTorch ambassador

Building AI agents with the Google Agent Development Kit. Survey of use cases: customer service, finance, supply chain, software development, cybersecurity.

IA Agentique : augmenter la confiance pour diminuer la surveillance

Raphael Antoine — healthcare interoperability expert

Real-world reflections after a year of daily AI agent use. Concrete takeaways, no dogma: where trust is earned, where supervision stays essential.

MCP Servers et Agents IA

A hands-on look at the Model Context Protocol (MCP) and how it connects AI agents to real data, services, and applications. How an agent talks to an MCP server to perform concrete actions.

Sandboxing pour Agents IA — du conteneur Docker à la MicroVM Firecracker

Your AI agent runs code, spawns shells, hits the network. How do you isolate what it shouldn't do? A tour of isolation boundaries: Docker, gVisor, Firecracker MicroVMs.

Introduction and Hands-on: AI Agents with Gemini SDK

Follow-up to the agents session (Jeremy Howard / Claudette). This time we apply the concepts with the Google Gemini SDK: 30 min intro to AI agents, then live coding.

Build an AI Agent — with code

English-speaking session. We follow the “How to Solve It with Code” course by Jeremy Howard and Johno Whitaker, reproducing the code live to understand how an AI agent works.

Build an LLM from Scratch

We follow Sebastian Raschka's “Build an LLM from Scratch” series. Bring your laptop: we build a large language model step by step, together.

Hands-on Session — Building GPT from Scratch! Part 2

Continuation of the interactive session following Andrej Karpathy's video. We pick up where we left off and keep building GPT.

Hands-on Session — Building GPT from Scratch!

Interactive session following Andrej Karpathy's iconic video: we build GPT from scratch, in Python, together.

Quantum Machine Learning / AlphaFold2 protein structure prediction

Bogdan Penkovsky (Alysophil) — plus AlphaFold2 talk

Double session: Quantum Machine Learning, myths and realities (by Alysophil R&D), followed by a presentation on protein structure prediction with AlphaFold2.