Agentic Financial Research Assistant
Finsense is a personalized financial research assistant that helps users make sense of market
uncertainty by combining live market data, news events, and risk analysis. Instead of giving
investment advice, it highlights what areas of the market are worth researching based on current
conditions and a user's preferences.
The system is built around an AI agent that reasons about information, supported by modular MCP
(Model Context Protocol) servers that provide reliable data and analytics. I developed three
specialized MCP servers: mcp_market processes market data to understand sector
and stock performance, providing historical returns and price summaries; mcp_news
retrieves relevant news headlines and extracts risk-related themes from unstructured text, mapping
these themes to affected sectors; and mcp_risk analyzes inter-sector relationships,
correlations, and volatility to identify how risk can spread across the market.
I implemented the entire backend infrastructure using FastAPI, integrating the MCP servers with
an intelligent coordinator agent that orchestrates multi-source financial analysis. The project
utilizes yfinance for real-time market data, RSS feed parsing for news aggregation, and custom
sentiment analysis algorithms. I also built a responsive web interface with JavaScript that
communicates with the backend API, deployed on Vercel with the API hosted on Render. The system
demonstrates advanced capabilities in asynchronous Python programming, API design, and creating
practical AI-powered financial tools.