Web Application
An AI driven finance platform that lets users query real time stock market data by company ticker, chat with financial PDFs, and browse company data through an admin dashboard. A multi-agent system built at a hackathon.
01Context
Built at a hackathon with a small team. I led frontend development and handled the database integration and API connection between frontend and backend.
02Problem
Investors and businesses had to jump between a stock ticker, financial filings and a search engine to answer one question about a company, with no single place to ask it directly.
03Approach
Route natural-language questions to specialized agents, one for live market data and one for document Q&A over financial PDFs, and surface both through one chat interface, with an admin view for the companies being tracked.
04Frontend
Built the ticker search, chat interface and admin dashboard in React, and wired them to the backend through a typed REST API.
05Backend & AI
Integrated the Yahoo Finance API for real-time pricing, and connected the frontend to a multi-agent backend so PDF-to-chat and ticker queries could share a single interface.
06Data
Chat history and company records persisted to Firebase, so a session's context carried across a user's follow-up questions.
07Challenges
Keeping the chat interface responsive when a query could resolve through either a fast live API call or a slower document-retrieval agent, without the UI treating them differently.
08Decisions
Kept the agent-routing decision on the backend and the frontend agent-agnostic, so adding a new agent type never required a client-side change.
Layer 01 / 4
React
Renders the ticker search, chat interface and admin dashboard, and dispatches every query to the agent layer through the REST API.
Key features
- Multi-agent AI architecture
- Real-time stock ticker queries
- PDF-to-chat
- Chat history
- Admin dashboard
- Yahoo Finance API integration
Result
Built end-to-end at a hackathon as a working demo, from live ticker data to PDF chat to an admin dashboard. Part of a hackathon track record that includes multiple first-place finishes.
What I'd improve
Add caching for repeated ticker queries, and stream agent responses token-by-token instead of waiting for the full answer.
Next dossier · 02
PDF ChatBot