Startup Analyser
AI startup analyser using RAG system
01 / Overview
Startup Analyser provides deep, actionable intelligence on emerging companies by leveraging a sophisticated Retrieval-Augmented Generation (RAG) system. Investors and analysts can explore massive datasets of startup metrics, pitch decks, and market positioning. The system combines Redis caching and MongoDB for extreme speed, presented via a sleek React interface.
03 / The Hardest Path
Achieving accurate context retrieval from highly unstructured data like pitch decks and complex financial metrics.
04 / Challenges
Standard RAG approaches kept returning irrelevant chunks of text. Furthermore, querying the vector database simultaneously with MongoDB caused high latency on the frontend.
05 / Solutions
I tuned the RAG embedding strategy to prioritize financial semantics and structural context. To solve the speed issue, I implemented Redis to cache frequent queries, slashing load times for the React UI.