summary
10+ years laying the groundwork for ML systems that survive real load. Worked across NLP, computer vision, time-series and generative AI. I'll write the model, design the infra, lead the team, or call your idea stupid before it costs you €750k... whatever your project actually needs to go live.
i'll make sure your project goes live. no buzzword bingo, no selling you breathless hype for the sake of it.
selected engagements
Machine Learning Engineer@Fibercop / BeExact
04/2025 → 03/2026
KKR-funded fibre rollout, Italy. Architecture, prototyping and full rollout.
Built a computer-vision platform processing petabyte-scale RGB panoramas + LIDAR data, to help fix the 'last mile' problem in installing fibre optic internet, replacing manual on-site inspections. Yearly savings: €3–5M. First successful ML deployment for a traditional Java software shop; now their stepping stone toward broader AI capability.
stack · multimodal LLMs · YOLO · self-hosted k8s · Python · Postgres
AI Engineer@BMW Group
04/2024 → 03/2025
Financial Services division. Productionising generative AI for thousands of users.
Productionised a text-to-SQL agent serving 9k B2B users, who use it to build dashboards over financial data through chat. Built lightweight LangChain-alternative primitives and an n8n-style orchestrator from scratch, all focused on low-latency, auditable, secure GenAI on enterprise AWS... Zero LangChain spaghetti, by design.
stack · AWS · Azure OpenAI · Go · Python · Postgres + pgvector · ECS
Machine Learning Engineer@Port of Rotterdam
2022 → 2024
Computer vision at port-scale. Full lifecycle ownership.
Built an object-detection pipeline using camera feeds from port inspection vehicles on Azure. Owned cloud architecture, edge connectivity and hardware design, MLOps maturity, business-case validation, and integration. Learnings allowed Port of Rotterdam to expand their computer vision ambitions from roadside inspection to drone and maritime video analysis.
stack · Azure · k8s · Python · Postgres / PostGIS · YOLO · OpenCV
Senior Consultant, Data Science & Engineering@Eneco
2018 → 2022
Centre of Excellence: raising data maturity across an energy major.
Designed and shipped a self-service data platform on Snowflake + TimescaleDB serving four business units. Built a deep-learning forecasting service still in production today: solar yield for 200k households, wind output for 150+ wind parks. Spearheaded the AI-driven greenhouse control system that spun off into BlueRadix. Also led a team modelling B2C churn, CLV and acquisition-cost simulations.
stack · Snowflake · AKS · Python · AWS · TimescaleDB · MLOps
Senior Data Scientist@Capgemini
2015 → 2018
Notable client work: KLM, Dutch National Police
At KLM Flying Blue: recommender systems and personalised marketing; replaced an 8h SAP low-code pipeline with 300 lines of SQLite (24× faster).
At the Dutch Police: high-throughput tool for enrichment, classification and migration of internal mail items, 30× pipeline speedup.
independent work & side projects
Smokescanner · computer vision system monitoring Tata Steel emissions at Wijk aan Zee. 28M frames analysed on an NVIDIA Jetson; hundreds of emission events of toxic dust clouds verified.
YOLO · Postgres · edge computing
SimpleSSTV · native iOS app for slow-scan television (ham-radio image transmission). Profitable, organic users, zero ads.
App Store · Swift
Pagina100 · Apple TV app for browsing Dutch teletekst. Yes, teletekst is still alive in 2026.
tvOS · Swift
PD3BAL · licensed amateur radio operator (NL).
occasionally pinging the ionosphere
tech & tooling
languagesPython, Go, Bash, SQL, R, Swift
infraKubernetes, Docker, AWS, Azure, bare-metal, serverless GPU
dataPostgres (pgvector, PostGIS, TimescaleDB), Snowflake, Databricks, DuckDB, Parquet, dbt
ml & aiPyTorch, scikit-learn, OpenCV, YOLO, OpenAI / Azure OpenAI, MLflow, OpenWebUI, Claude Code
opsPrometheus, Grafana, FastAPI, n8n, NVIDIA / CUDA, linear & numerical optimisation
education
PhD, Computational Chemistry · VU Amsterdam 2015
·
MSc & BSc, Chemistry 2011 · 2009
~$ whoami
Living in Delft with wife, three kids and a cat. Cooks, reads sci-fi, listens to techno, tinkers with 3D printers and synthesizers.