2026 · AI / ML / Data Engineering
Census Income MLOps API
An income classifier behind a FastAPI service with slice metrics, CI, and continuous deployment.
- Census data
- Income classifier
- FastAPI inference
Complete course-style projectGitHub ↗
Problem
Take a model from raw census data to a tested, deployed inference API with documented behaviour.
My role
Author of the repository.
Approach
- Clean the UCI Census Income data (32,561 rows), train a RandomForest with a stratified 80/20 split.
- Compute per-category slice metrics; document the model in a model card.
- Serve via FastAPI (/predict), test with pytest, GitHub Actions CI, auto-deploy to Render.
Results
- Held-out test set: precision 0.8083, recall 0.5727, F1 0.6704 (model card).
Stack
- Python
- scikit-learn
- FastAPI
- GitHub Actions
- Render