AI engineer & data scientist
Portfolio · Blog · CV · LinkedIn
AI engineer and data scientist working end-to-end, from data and pipelines to models and finished tools. Currently working on LLM embeddings, forecasting and data platforms.
My background is in Python/SQL pipelines, data validation and reporting datasets for energy, renewables and agribusiness. Working with operational, environmental and GPS data taught me to investigate the inputs behind a result.
Based in Berlin.
I built Guardrails in LLM Deployment to compare baseline RAG with layered guardrails. The learning assistant combines BGE-M3 retrieval, Chroma metadata filtering and OpenAI-compatible classification, answer generation and verification.
You can try the offline demo, inspect the automated tests and use local tools to review evaluation data. The workflow separates calibration from final evaluation.
Explore the code · Try the offline demo · Read the evaluation notes
Where should a charging operator expand next? I built NRW Charging Atlas to support early siting decisions for operators planning charging infrastructure for electric vehicles (EV) across all 53 districts in North Rhine-Westphalia (NRW), Germany.
You can screen districts for a rollout, inspect what drives their priority rankings and test proposed stations against current coverage. I pushed spatial joins and scoring into PostGIS, with Python for data validation and loading, GeoServer for map and feature services, and GeoNode for dataset discovery and metadata.
Explore the code · Read the project story · Try the visual walkthrough · Open the slides
Look inside the GeoNode dataset catalogue
The catalogue keeps published layers and their source information together. Proposed stations start empty; district layers share boundaries, so their previews can look alike even though their values differ.
Map previews: © EuroGeographics and © OpenStreetMap contributors.
agri-weather-yield-drivers combines weather, soil and yield data into explainable risk signals. The work includes checks on spatial joins, coverage and baseline choices, with a reporting mart and risk-zone GeoJSON as outputs.
bi-python-uv-project-scaffolder handles the initial structure and setup for BI projects using Python and uv, so I can start working on the pipeline.
- NRW Charging Atlas: comparing charging coverage with GeoNode and PostGIS: the siting question, scoring model and data pipeline.
- A quick tour of NRW Charging Atlas: start the application, try a proposed station and explore the data services.
- Crude benchmarks: API, sulfur and interactive charts: explore crude quality through interactive charts.
Python · SQL · RAG · LLM Evaluation · BGE-M3 · Chroma · PostgreSQL / PostGIS · DuckDB · GeoNode · GeoServer · pandas · NumPy · scikit-learn · PyTorch · Power BI · Docker · AWS · TypeScript · React
I'm looking for AI Engineer roles focused on Python applications, retrieval and model evaluation. I'm also interested in data engineering within AI teams. Find me on LinkedIn or explore my portfolio.




