AI & ML · Research prototype
AeroOps ML
An aviation ML pipeline for flight-delay prediction and engine remaining useful life, with data-quality checks and model serving.
PythonPySparkXGBoostMLflow
Explore repository 
The project
A PySpark bronze, silver, and gold pipeline over public flight data, paired with turbofan remaining-life experiments. Feature-group ablations and MLflow tracking support time-based flight validation and engine-group validation. FastAPI exposes XGBoost delay and remaining-life predictions, while a Streamlit dashboard presents KPIs and model results. An LSTM is included in the research comparison rather than the serving path. Cloud Run and App Runner files are deployment configurations, not evidence of a hosted service.
Repository activity ·
Inside the build
- 01PySpark data-quality gates, partitioned Parquet, and a DuckDB KPI layer
- 02Time-based flight-delay validation and engine-group RUL evaluation
- 03MLflow experiment tracking, feature ablations, and SHAP explanations
- 04FastAPI prediction endpoints and a Streamlit results dashboard