Skip to content

GeneralR3d/TechFin

 
 

Repository files navigation

TechFin — Know the World Before the Market Does

Real-time geopolitical intelligence meets personal portfolio management.
Update: 1st place winner of the 2026 NTU Fintech Innovators Hackathon

TechFin is an AI-powered personal finance dashboard that doesn't just track your stocks — it reads the world. By aggregating news across multiple sources, performing deep sentiment analysis, and mapping relationships between global events and financial sectors, TechFin gives retail investors the kind of macro-awareness that used to be reserved for institutional traders.

Demo

Try it live at tech-fin-xi.vercel.app and log in with:

  • Username: demo
  • Password: password

Demo Video

Demo Video


The Problem

Most retail investors react to the market. TechFin helps you get ahead of it.

TechFin is world-event-focused. It identifies like most mentioned geopolitical world events, mergers and acquisitions, as well as government macro-policy announcements to find the the most up to date investor sentiments related to the affected sectors, and quickly makes investment suggestions.

What Makes TechFin Different

At its core, TechFin uses a custom knowledge graph (powered by Neo4j) to model the relationships between news articles, world events, sectors, and tickers. This enables intelligent graph traversals that surface the most contextually relevant insights for each event — not just the loudest headlines, but the most connected ones.

On top of that graph sits a LangGraph multi-agent AI chatbot that can answer natural-language questions about your portfolio, current market themes, and recent events by querying the knowledge graph in real time. Ask it "What's happening with NVDA?" and it pulls live, structured data — not hallucinations.


Neo4J graph


Core Features

Feature Description
Event Intelligence Detects top geopolitical events, M&A, and macro-policy shifts in real time
Sentiment Grid Sector-by-sector sentiment heatmap updated daily
Graph-Powered Matching Neo4j graph traversals match news to the most affected sectors and tickers
AI Chatbot LangGraph multi-agent assistant that queries your knowledge graph to answer market questions
Portfolio Recommendations LLM-powered stock suggestions tuned to your holdings
Multi-Source Aggregation Yahoo Finance, Reddit, Twitter, LinkedIn — all in one feed
Social Pulse Community sentiment from Reddit and social platforms per ticker

Tech Stack

Frontend — React 19 · Vite 6 · TypeScript · Tailwind CSS 4 · shadcn/ui · SWR

Backend — FastAPI · SQLAlchemy 2 (async) · PostgreSQL · Alembic

Intelligence Layer — Neo4j (graph traversal) · LangGraph + LangChain · OpenAI gpt-4o-mini · RapidAPI (Yahoo Finance) · Reddit OAuth · APScheduler (hourly ingestion)


Team

Builder Handle
Tuan Ding Ren @General3d
Jayce Goh @JyG12
Boxuan @boxuan-yu
Shianne @zikyuu
Jden Goh @jdengoh

Getting Started

Prerequisites

  • Node.js ≥ 18
  • Python ≥ 3.12 + uv
  • Docker (for PostgreSQL + Neo4j)

Infrastructure

docker compose -f docker-compose.infra.yaml up -d

Backend

cd backend
uv sync
cp .env.example .env        # edit DATABASE_URL and API keys as needed
uv run alembic upgrade head
uv run python scripts/seed.py
uv run uvicorn app.main:app --reload --port 8000

Frontend

cd frontend
npm install
npm run dev                 # dev server on localhost:5173

The Vite dev server proxies all /api/* requests to the FastAPI backend on :8000.

Or run everything at once: make dev


Environment Variables

Copy backend/.env.example to backend/.env and fill in:

DATABASE_URL=postgresql+psycopg://USER:PASSWORD@localhost:5432/techfin
SECRET_KEY=change-me-in-production
NEO4J_URI=bolt://localhost:7687
NEO4J_USERNAME=neo4j
NEO4J_PASSWORD=your_neo4j_password
OPENAI_API_KEY=your_openai_key

OPENAI_API_KEY is required for the AI chatbot and graph entity extraction.


Demo

Try it live at tech-fin-xi.vercel.app or run locally and log in with:

  • Username: demo
  • Password: password

About

A market intelligence platform transforming global news, social chatter, and portfolio context into actionable investing suggestions for users. At its core, Neo4j knowledge graph maps events, entities, sectors, and tickers so system can traverse relationships, surface second-order market impacts. Real-time queries power LangGraph agentic assistant

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages

  • TypeScript 52.2%
  • Python 46.0%
  • CSS 1.4%
  • Mako 0.2%
  • Shell 0.1%
  • HTML 0.1%