Principal Engineer · Production AI Systems
16 years in production. Today I build AI systems: agents, retrieval and context systems, and integrations with the tools businesses already use. I take them from architecture to production.
AgentRH · Partner, Founding AI Engineer
Building the client-context layer of AgentRH, an HR copilot inside Microsoft Teams. Context engineering, RAG on company knowledge bases, workflow automation, and connectors to SharePoint, Teams and existing systems. Each company’s policies, organisation and tools provide the context for its answers.
Memoire: local-first governed memory for AI coding agents. Capture and retrieve context across coding tools in a store you own. AGPL-3.0.
Safe: tooling for safer package execution, dependency auditing and guarded installs. MIT.
Contributions: Laravel & Ecosystem, Vue/Nuxt Ecosystem, Paperclip, Uptime Kuma.
| Work | Result |
|---|---|
| Tourism marketplace, 8 years as sole technical lead | 3× deliverables for the same budget |
| National real-estate portal, lead developer / architect | 5× deliverables for the same budget |
| Image serving | Costs divided by 15 |
| Kubernetes infrastructure, across multiple clusters | Bills halved |
- AI systems: context engineering, RAG, agents, model fine-tuning and business integrations.
- Application delivery: Laravel, Vue.js/Nuxt and TypeScript, with PostgreSQL and Elasticsearch.
- Orchestration and operations: Kubernetes, Temporal, Kafka and CI/CD.
- Own AI infrastructure: 10 GPUs, 240 GB VRAM under Kubernetes for local inference and fine-tuning. Cloud infrastructure when the project calls for it.
Standalone 2–5-day audits, scoped projects and fixed-fee delivery. Available 3 days/week or for fixed-fee deliverables, remote from France, serving clients in Europe.
- Business: tsbwc.com · contact@tsbwc.com
- Open source: michel@superbiche.me
- LinkedIn: micheltomas
- Malt: micheltomas





