Selected work
and experiments
Multi-agent orchestration platform I co-founded. 5+ AI agents running in parallel with shared memory. Slashed token costs by 80% with intelligent context routing.
Distributed orchestrator for AI workers (Claude, OpenAI, Gemini). Real-time FastAPI API, Redis broker, scalable workers.
AI content engine I built as CTO. Learns your writing voice, adapts to your niche. Adaptive knowledge base, central API, and continuous training pipelines.
Growth Operator. Go-to-market of a desktop app for testing mobile and web apps: visual test flows, simulators, physical devices and a cloud device farm.
My customer acquisition agency. Funnels, landing pages, CRMs and internal tools shipped to clients, with data isolated per client. Next.js, Firebase, Vercel, Stripe.
Tools I publish
for agent builders
Long-term memory for coding agents, done with files you can read. A blueprint and a tested MCP server: Markdown notes, ranking by relevance, confidence and freshness, and superseding instead of overwriting.
A knowledge base your AI agents can actually use. Method, Obsidian-compatible vault template, and a linter that keeps notes findable, linked and current.
Experience
& education
Agents in production: water plant optimization, AI tender analysis with cited sources, a Claude coding agent and an Anthropic API proxy on the group's secure LLM, Gen AI infrastructure on Google Cloud.
Procurement tool for water treatment projects, from projects to tenders to purchase orders, with GitLab CI/CD to Google Cloud.
Automated lab certificates and after-sales workflows.
Building the orchestration layer for AI agents. Multi-agent platform with shared memory and parallel execution.
Built an AI-powered SaaS from zero. Content engine with adaptive training pipelines and real-time personalization.
Software & Information Systems. Pepite France, national student-entrepreneur program.
Ideas, builds
& lessons learned