Skip to content

ridash2005/AlphaStream-Presentation

Repository files navigation

AlphaStream India 🇮🇳

React Google Gemini Pathway DuckDB FastAPI LangGraph

AlphaStream India is a professional-grade, real-time financial intelligence terminal designed specifically for the Indian retail investor. It bridges the massive information gap between institutional traders and retail participants by providing deterministic, agentic, and sub-second market insights.


🚀 Key Features

  • 13-Agent Reasoning Swarm: Powered by Gemini 2.0 Flash, our swarm fuses technical trends, corporate filings, insider trades, and news sentiment into a single, actionable Alpha Score (0-100).
  • Sub-2s Latency: Built on Pathway’s streaming engine, the system ingests data from 14+ connectors (NSE, BSE, NSDL, Groww, FRED) and pushes live signals to the UI in under 2 seconds.
  • Deterministic Analytics: Unlike generic chatbots, AlphaStream uses a Text2SQL pipeline executing on DuckDB for precise, hallucination-free numerical answers.
  • Regulatory Moat: Integrates India-specific regulatory data (NSE SAST/PIT, NSDL FII/DII) that global platforms often lack.
  • Adaptive RAG: Implements geometric document retrieval to optimize token costs by 40% while maintaining rigorous context for complex queries.

🛠️ Technology Stack

Layer Technology Purpose
Frontend React 19, Vanilla CSS Real-time terminal UI with WebSocket streaming
AI Orchestration LangGraph, Gemini 2.0 Flash 8-node NLQ pipeline with self-correction loops
Streaming Engine Pathway Incremental indexing and real-time data ingestion
OLAP Database DuckDB Deterministic financial analytics and Star Schema
Vector Database ChromaDB High-performance vector embeddings for RAG
Online ML River (ML) Streaming anomaly detection (Z-Score, ADWIN, HST)
Backend FastAPI, Python High-performance async API endpoints
Visuals Manim (Community) Dynamic mathematical animations for presentation

👨‍💻 Author

Rickarya Das
AI & Financial Systems Architect


📈 Impact

  • Productivity: Recovers ~₹2.19L of lost productivity per user/year by compressing 2 hours of research into 15 minutes.
  • Alpha: Delivers 2-3 actionable signals monthly with 60%+ historical accuracy.
  • Risk: Reduces behavioral panic selling via portfolio-aware alerts and FII/DII streak tracking.

🔗 Project Links

  • Repository: AlphaStream India
  • Problem Statement: PS-06 · AI for the Indian Investor
  • Event: ET GenAI Hackathon Finale 2026

GitHub Topics

fintech, generative-ai, multi-agent-systems, streaming-data, financial-analysis, react, fastapi, duckdb, pathway, gemini-ai, quantitative-finance, india-market

About

A multi-agent, streaming financial intelligence terminal for the Indian retail investor. Built with a 13-agent reasoning swarm (Gemini 2.0 Flash) and Pathway's streaming engine, it delivers deterministic, sub-2s market insights by fusing technical trends, corporate filings, and India-specific regulatory data (NSE/NSDL).

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors