Chat With Your Data
Designed and deployed a scalable multimodal RAG platform handling 1TB+ of text, documents, images and structured data. Built two REST APIs for automated ingestion and AI-powered querying, combining vector search with fine-tuned LLM pipelines to deliver real-time, context-aware answers and cut manual analysis by 80%+.
A retrieval platform for organisations sitting on terabytes of unstructured knowledge.
The system ingests text, PDFs, images and structured records, normalises them into a unified vector index, and serves answers through a query API. Ingestion and querying are separated into two services so each can scale independently. Retrieval is tuned with hybrid search and reranking, and responses are grounded with citations back to the source documents.
Tools: LLMs, RAG, LangChain, LlamaIndex, Chroma, prompt engineering, text embedding, NLP.