How fauziiifau/qc_ai_service works
This is a small FastAPI defect-recommendation service. A client submits a process and defect type; the endpoint validates them, builds the model input, and returns the model’s recommendation. The source shows the model artifact being loaded at startup and invoked for prediction; its internal structure is not visible, and the README is empty.
Main entry points: main.py Prediction endpoint · model_defect_v2.pkl Trained model
Components 7
Other components
- API client
- Prediction endpoint
main.py - Input validation
main.py - Input dataframe
main.py - Model pipeline
main.py - Trained model
model_defect_v2.pkl - Recommendation response
main.py
Connections 6
- API client → Prediction endpoint: submits request
- Prediction endpoint → Input validation: validates input
- Prediction endpoint → Input dataframe: builds input
- Prediction endpoint → Model pipeline: predicts
- Prediction endpoint → Recommendation response: returns recommendation
- Model pipeline → Trained model: loads at startup