DATA ENGINEERING & ML · 2025 to 2026
F1Dash
Personal projectFrom race telemetry to predictions.
Feature preparationVenue history · Driver & team performance · Weather · Rolling form
Pipeline outputsDuckDB feature tables · Saved model artifacts · Race-outcome predictions
THE PROBLEM
What needed solving.
Race telemetry, weather, and historical results need to be prepared together before they can support race-outcome predictions.
What I implemented.
Data pipeline & machine learning development
- Built Python ingestion and transformation scripts that materialize race, lap, and weather data into Bronze, Silver, and Gold DuckDB schemas.
- Aggregated venue history, driver and team performance, weather conditions, and rolling form into model-ready feature tables.
- Implemented Scikit-learn training and prediction scripts, saving model artifacts and race-outcome predictions for downstream use.
- Connected stored predictions to API endpoints and a React dashboard for exploring race outcomes and supporting statistics.
ENGINEERING DECISIONS
The technical challenge.
The technical challenge was turning different race inputs into consistent model features. The pipeline materializes Bronze, Silver, and Gold DuckDB schemas before training, then saves models and predictions for API and dashboard access.
THE RESULT
What changed.
A complete path from source telemetry to model-ready data and stored predictions. The public repository includes pipeline documentation, architecture notes, and setup instructions.
Explore the repository