Hi I'm Jeel. I turn messy data and raw pixels into systems that actually run: models, the pipelines that feed them, and the APIs that serve them. Hand me twenty years of chaotic Excel exports or a camera pointed at something hard to read, and you'll get back a model with a CLI, a Docker image, and documentation. Happiest where tabular ML, computer vision, and optimization overlap. I work in German and English.
| Project | What it does |
|---|---|
| billet-stamp-OCR | Reads stamped ID numbers off steel billets — computer vision on a real factory floor |
| shoe-recognition-jeel-project | CLIP-based visual search: give it a photo, it finds look-alike shoes |
| Clustering_project_jeel | Segments customers by behavior with classic clustering |
| stock-api-jeel-project | A Dockerized FastAPI service that serves live stock data |
Off the data-science path, two builds I keep around: LiFi-Based-Transmission, an Arduino link that sends text through an LED bulb and reads it back with a solar panel, and weather-data-app-jeel, a Java OOP exercise that generates a static weather site.
Python for everything. scikit-learn and XGBoost when the data is tabular, PyTorch and OpenCV when it's pixels, FastAPI and Docker when it needs to ship.
Regenerated every night by GitHub Actions.
