At Tata Consultancy Services, I build Python-based AI applications, production APIs, demand forecasting solutions, and scalable data workflows. My forecasting work using XGBoost and Prophet achieved 95% daily forecast accuracy.
I've developed 5+ FastAPI APIs for LLM-powered applications, served low-latency models with vLLM, and fine-tuned a LLaMA 3.1 14B model for insurance claims recommendations. That work improved LLM accuracy by 35%, reduced manual effort by 50%, and delivered production inference latency below one second.
I also process large-scale data with PySpark, Databricks, and AWS, reducing forecasting compute time by 81%. I enjoy building reproducible ML and GenAI systems with MLflow, Docker, GitHub Actions, RAG, and reliable deployment practices.
