Akshay Shewatkar

• Built a RAG pipeline (LangChain + LlamaIndex) with semantic chunking, entity ex- traction, and similarity search for structured PDF extraction, validated via Pydan- tic. • Developed an LLM evaluation framework (LangGraph) to benchmark hallucina- tion rate, latency, and cost across models, with stakeholder-facing reporting. • Integrated Celery, Redis, and Flower into an existing Django codebase for async task processing; wrote and improved DRF endpoints with Swagger documenta- tion.