MbarQ (Belgium)
Apr 2025 – PresentAgentic Engineering Lead
May 2026 – Present- Lead MbarQ's agentic engineering practice — setting the standards, tooling, and enablement for building with AI coding agents company-wide.
- Defined the team's agentic way of working and drove company-wide adoption through internal workshops, reusable tooling, and industry research.
Agentic Engineering Knowledge Base & LLM-Wiki — Team-wide Claude Code standards, on-demand doctrine, and a library of reusable skills and playbooks, paired with an "LLM-Wiki" knowledge vault that agents maintain and that mirrors the internal repos. Tech: Claude Code, Markdown, Python, Git hooks / CI.
Claude Code Teams Bot — A Microsoft Teams bot that runs Claude Code autonomously: a queue-backed worker executes coding jobs on Azure and posts progress cards back into the chat, gated by an AAD sender allowlist and a secrets sandbox. Tech: TypeScript, Microsoft Teams SDK, Azure Container Apps, Bicep.
AI Translator Workflow App — A cross-platform desktop app that wraps Claude Code in headless mode to guide non-technical "AI Translators" through a structured, multi-stage project-scoping workflow, writing artifacts straight into the shared team vault. Tech: Python, FastAPI, React, pywebview, PyInstaller.
Agentic Landscape Site — A static microsite that maps MbarQ's entire agentic way of working on one screen — pillars, agent layer, and role-based onboarding paths — and distributes the team's desktop apps. Content is federated from four repos at build time, guarded by link-check gates, a doc-drift tripwire, and post-deploy smoke tests. Tech: HTML/CSS/vanilla JS, Python, GitHub Actions, Azure Static Web Apps.
Senior AI Architect & Engineer
Apr 2025 – May 2026- Led end-to-end GenAI engagements: discovery, architecture, hands-on implementation, and go-live support.
LaunchPad (Internal) — Built a modular RAG accelerator framework on Azure for enterprise chatbot deployments; extended it with an agentic architecture to automate multi-step workflows. Tech: Python, FastAPI, Azure OpenAI, LangChain, Vue 3.
Fiducia 4.0 (Puratos) — Predictive quality control for food manufacturing: ML classifiers that predict QC failures of finished products at run level, before lab results come in, combining plant process/sensor (PI) data, raw-material lineage, and historical QC measurements. Built the production-grade pipeline end-to-end on Azure ML — automated batch detection and feature engineering, per-product/per-criterion model training, and drift monitoring — deployed across multiple production sites with differing data availability. "Golden recipe" analysis translates the models into optimal process settings and diagnostics operators can act on, linking predictions to concrete quality and cost improvements. Tech: Python, Azure ML.