How sifoukhemira/mlops-tp works
This is a small fraud-classification service with a model-training workflow and a FastAPI prediction API. Training reads a fraud dataset, prepares features, fits and evaluates a classifier, and saves model/schema artifacts while logging the run to MLflow. The API loads those artifacts, accepts schema-validated requests, and returns predictions and probabilities. The README is empty; the graph reflects the shown source excerpts, so it does not infer behavior for the empty inference or config modules.
Main entry points: api.py FastAPI service · train.py Train workflow · artifacts Model artifacts · schemas.py Prediction request
Components 7
Model training
- Train workflow
src/mlops_tp/train.py
Prediction API
- FastAPI service
src/mlops_tp/api.py - Prediction request
src/mlops_tp/schemas.py
Model artifacts
- Model artifacts
src/mlops_tp/artifacts
Other components
- Fraud dataset
- MLflow tracking
- API client
Connections 7
- Train workflow → Fraud dataset: reads
src/mlops_tp/train.py - Train workflow → Model artifacts: writes
src/mlops_tp/train.py - Train workflow → MLflow tracking: logs run
src/mlops_tp/train.py - API client → FastAPI service: requests prediction
- FastAPI service → Prediction request: validates with
src/mlops_tp/api.py - FastAPI service → Model artifacts: loads
src/mlops_tp/api.py - FastAPI service → API client: returns prediction
src/mlops_tp/api.py