Sumedh Bhatwadekar

At Capgemini, I develop machine learning solutions for churn prediction, anomaly detection, and customer segmentation using enterprise datasets. I work across model training and validation, and use MLflow and DVC to support experiment tracking and reproducible workflows.

I also build Dockerized inference services with FastAPI and REST APIs, and run training workflows on AWS EC2 and SageMaker. My work includes supporting deployment pipelines, CI/CD automation, and production monitoring for internal ML systems.

For my M.Tech AI/ML dissertation, I designed an enterprise knowledge assistant using hybrid RAG, combining FAISS semantic retrieval with BM25 and RRF. The project also covers MCP-based context continuity and LLMOps workflows for tracking, deployment, monitoring, and evaluation.