I build practical machine learning systems for real-world, messy data problems.
A data science/ML practitioner with strong engineering instincts. I focus on messy, real-world data and applied modeling (not just theory). I care about bias, uncertainty, validity, deployment, and usability. I've worked on supply chain optimization, ecological/foraging prediction, and full-stack systems.
📈 20% cost reduction in delivery logistics
Multi-tenant web application for optimizing local food distribution logistics. Designed cost allocation models (fuel, depreciation, labor). Integrated database + backend (Flask + MySQL). Worked with real stakeholders (OGC).
🎯 89% accuracy in species prediction
Built models using iNaturalist + WorldClim data. Explored embeddings (Word2Vec-style) for ecological relationships. Focused on predicting species presence in microclimates. Addressed noisy, biased observational data.
âš¡ 99.9% uptime across all services
Deployed multiple services using Docker + Nginx. Managed databases and APIs on VPS. Built and hosted production-style apps with monitoring.
🚀 3x faster experiment iteration
Built reusable Python package for A/B testing with statistical hypothesis testing. Implemented common ML metrics (RMSE, MAE, R², F1). Created visualization dashboard for experiment results.
AWS (EC2, S3, SageMaker), Docker, Linux, Git
Production deployments on AWS
PyTorch, TensorFlow, Scikit-learn, Pandas
15+ models developed
MLflow, Model versioning, Performance tracking
Real-time model monitoring
Statistical hypothesis testing, Metrics design
200+ experiments analyzed
Spark, Distributed computing, Data pipelines
Processing 100GB+ datasets
Optimization, Embeddings, Clustering, Recommender systems
Mathematical foundations