End-to-end systems, analytics, and ML pipelines
A selection of practical, production-minded work with clear outcomes and thoughtful delivery.
From data ingestion to deployment, with clear documentation and handoff.
Practical models aligned to real operational decisions and constraints.
Transparent outputs, evaluation reports, and stakeholder-friendly artifacts.
Two deep dives showcasing problem framing, pipeline design, and delivery-ready outputs.
Forecasting
Retail demand forecasting to improve replenishment accuracy and planning speed.
Daily demand forecasts by store and SKU to reduce stockouts and overstock.
Time-series pipeline with calendar, promotions, and weather features. MLflow tracking and model selection on rolling validation.
FastAPI endpoints with a Streamlit UI, plus evaluation reports by region and category.
Risk Scoring
Risk scoring pipeline to support consistent underwriting decisions.
Predict probability of default while maintaining fair, explainable approvals.
Modular pipeline with feature engineering, model comparison, and calibration checks.
Risk scoring API with a demo UI and CI checks for model stability.
Portfolio
Additional projects across analytics, BI, and quantitative research.
Forecasting pipeline with evaluation reports and a production-ready API.
Reusable preprocessing and prediction APIs for retention analytics.
Risk scoring pipeline with explainability and CI support.
Interactive KPIs and executive reporting views.
CTE analysis to surface trends and risk factors.
Risk-aware optimization with benchmarking and drawdown analysis.
See more on GitHub.
I can walk through architecture choices, metrics, and deployment workflows.