A quick index of all micro-tasks in this repository. Each task is a self-contained lab with its own README, LEARNING guide, code, and docker-compose environment.
- Minimal Fastify HTTP service
- Health and info endpoints
- Fastify service with
/predictendpoint - JSON schema validation
- Refactored ML layer module
- Separation of concerns
- TF-IDF + Logistic Regression
- Introduces basic ML pipeline
- train/test split, accuracy metrics
- Text cleaning preprocessor
- Lowercasing, punctuation removal, stopwords
- k-fold cross-validation
- Model evaluation improvements
- GridSearchCV for hyperparameter optimization
- best_params, best_cv_accuracy
- Semantic version folders (1.0.0, 1.1.0…)
- Saved metadata.json
- Model versioning system
- Structured registry folder for model artifacts
- ModelRegistry abstraction
- Python Model API (FastAPI)
/predictand/healthendpoints- Model registry integration
- Dockerfile + containerised inference
- Dockerized FastAPI service
- Full stack running via docker-compose
- Fastify gateway + Python API
- requestId propagation Fastify → Python
- Structured logging across services
X-API-Keyenforcement for protected endpoints- API key authentication layer
- Docker volume for models
- Python auto-discovers versions
- Fastify exposes
/admin/models&/admin/models/latest
Each micro-task can be run independently:
cd "<micro-task-folder>"
docker compose up --build