Senior Data Scientist at Effectory. Eight years shipping search, NLP, forecasting and LLM systems.
01Work
Now
Effectory · Amsterdam
2023–2026 · Contract
Google · San Jose, CA
Built production document pipelines with Cloud Composer, Document AI and Gemini to turn thousands of unstructured PDFs into BigQuery datasets. Developed a Gemini Text-to-SQL agent with Vertex AI vector search and a Flask API, and added monitoring for LLM quality, cost and latency.
2017–2023
Thermo Fisher Scientific · Carlsbad, CA
Built production ML for service and customer teams, including BERT-based ticket classification, semantic search over support tickets, service-order forecasting, propensity models and recommenders. Used PySpark, Databricks and MLflow for data pipelines, deployment and reproducible experiments.
02Latest
I wanted a trip budget tracker that felt nice to open, so it’s styled as a boarding pass.
Set a budget, log what you spend, and share the trip with friends using an invite code. It started as a React web app. I’ve since rebuilt it natively for Android in Kotlin and Jetpack Compose and for iOS in SwiftUI, all on one Supabase backend. It supports 30 currencies at ECB rates and generates cover art for each trip.
03Side projects
More on GitHub.
04A bit about me
I’m a data scientist living in The Hague. For about eight years I’ve been building machine learning systems that run in production: semantic search, NLP, forecasting and, more recently, LLM pipelines.
Pipelines, deployment and monitoring matter to me as much as the model, because that’s how a model earns people’s trust.
Outside work I build side projects, mostly. I’m also learning Dutch, lifting, and taking photos with a Sony A7R IV.