Souritra Banerjee

Data Scientist @Amazon

Bengaluru, IN

Leading end-to-end ML and simulation systems for supply chain intelligence across Middle East, North America, and India operations.Selected impact: Architected THANOS — Monte Carlo simulation platform for network-level capacity planning. Modeled constrained/unconstrained scenarios for FC expansion, volume routing, and peak planning.~$2M+ projected annual optimization impact. Owned demand forecasting across 10+ regions — replaced time-series baselines with XGBoost/LightGBM ensembles. Improved accuracy 40%(MAPE: 5.2% → 3.1%), informing $450K+ inventory decisions. Led RAG-based operational intelligence system over 100K+ documents — context-aware retrieval replacing brittle rule-based lookups. Reduced error rate 40%. Evaluation framework adopted organization-wide. Built LLM-powered email classification & task routing system — automated triage reducing manual assignment time ~60%. Improved NLP classification recall 28% for supply chain event triaging (PyTorch/Hugging Face) — enabling automated handling of 10K+ weekly events. Applied causal inference across 200K+ sellers — 12% ROI lift,~$2M revenue impact. Implemented automated model monitoring for 8+ production models — 65% faster refresh cycle, 3× faster stakeholder response. Certified Amazon Bar Raiser — selected to conduct leadership bar assessment interviews across Science and Analytics hiring loops. Mentored junior data scientists on modeling best practices and experimentation design.Core stack: Python · XGBoost · LightGBM · PyTorch · TensorFlow · Hugging Face · LangChain · RAG · SageMaker · MLflow · AWS · SQL

B.P. Poddar Institute Of Management and Technology

Bachelor of Technology (B.Tech.)

Indian Institute of Technology, Madras

Master of Technology - MTech