C Lalmuansanga

Designed and implemented a cutting-edge Retrieval Augmented Generation (RAG) system that leverages a Knowledge Graph instead of a traditional Vector Database. Created a backend pipeline that can analyze and verify speech videos using an LLM and a SOTA vision model. Developed a Chatbot that can generate natural language responses based on information retrieved from various sources using Langchain agents and RAG. Worked on improving traditional RAG systems using Rerankers, RAG Fusion, Dynamic Chunking, Multi Query Retrieval. Successfully implemented Role Based Caching on RAG and Chatbot systems substantially reducing system cost. Deployed models in adherence to the latest LLMOps best practices.