ML/AI Engineer (training and evaluation pipelines for recommendation models)
Implemented and benchmarked multiple state-of-the-art recommendation models, including neural and transformer-based methods, from scratch. Built a complete ML pipeline covering data cleaning, feature engineering, model training, tuning, and evaluation for personalized recommendation tasks. Work also included deployment and production automation via Docker and CI/CD, which relies on consistently prepared training data. • Built a personalized hybrid recommendation system in PyTorch for repeat-purchase patterns and product discovery. • Implemented and benchmarked 5 recommendation approaches: neural collaborative filtering, transformer-based, graph-based, matrix factorization, and classical BPR. • Containerized models with Docker and set up CI/CD for FastAPI production deployment. • Produced an ICCS 2026 paper on personalized next-basket recommendation with interpretable cycle-aware purchase modeling.