AI-First Workspace with GDPR-Compliant RAG
ContextA Marseille AI company building an AI-first workspace: AI operatives grounded in the company’s own mail, calendar, channels, tasks and files. Role: Director of R&D, February 2025 to October 2026, team grown from 7 to 15.
Ship agents, retrieval and automation as one coherent product, on company data, with auditable actions and GDPR, ISO 27001 and SOC 2 expectations from enterprise prospects.
Defined and steered the R&D strategy: a scalable GDPR-compliant RAG on containerized Python services, a mutable vector index with proximity-aware compaction and LSH-Bloom routing, hybrid lexical-vector search with structured chunking, multi-agent orchestration, and a multi-tenant MCP execution sandbox (containers, RBAC, network policies, seccomp, tracing).
Median search time cut by more than half with recall above 95 %; orchestration measured above 99.5 % success with sub-3-second time to first token under load. Prix Data & IA (C’est la Tech BB, BFM Business) and Prix de la Transformation Numérique (SEPEM) in June 2025.
