PORTFOLIO DEMONSTRATION PROJECTv2.4.0

EdgeIQ — Next-Gen AI Market Intelligence & Smart Trading Journal

Designed & Built by Raj Puthawala

EdgeIQ is a personal AI-powered financial market intelligence and trading-journal application. It combines near-real-time market data from Finnhub, financial-news sentiment classification using HuggingFace FinBERT, and AI-assisted qualitative research powered by Groq-hosted Llama 3.3 70B.

GitHub Repository Live Demo

Engineering Purpose & Competency Demonstration

EdgeIQ was built by Raj Puthawala to demonstrate full-stack engineering proficiency across modern enterprise application domains:

01. NEXT.JS 14 APP ROUTERServerless Route Handlers, SSR/Client boundaries, state hydration
02. LLM & FINANCIAL NLPStructured Groq/Llama JSON schemas, HuggingFace FinBERT sentiment pipeline
03. ARCHITECTURE & SECURITYUpstash Redis rate limiting, Zod schema validation, untrusted data isolation

Technology Stack

Frontend

Next.js 14 App Router

React, TypeScript, Tailwind CSS, Recharts, Framer Motion

Backend & API Layer

24 Serverless Routes

Zod Validation, Upstash Redis Rate Limiting

AI & NLP Engines

Groq / Llama 3.3 70B

HuggingFace FinBERT NLP (ProsusAI/finbert)

Market Data & Hosting

Finnhub Market Data

Vercel Edge Deployment

AI Security & Data Pipeline Architecture

User InterfaceNext.js API HandlersUntrusted Data BoundaryGroq Llama 3.3 + FinBERTValidated Zod JSON

All external news, user trade theses, and unstructured text inputs pass through untrusted-data boundaries (=== EXTERNAL DATA ===) with instruction isolation to protect against prompt injection and unsafe model outputs.

Application Limitations

•Journal Persistence: Uses browser localStorage persistence for the trading journal log. There is no cloud database or user login authentication system implemented.
•AI Qualitative Context: AI research analysis is powered by Llama 3.3 70B via Groq and reflects qualitative context based on provided snapshots.
•Market Observations: AI Market Observations are qualitative text patterns generated from financial context, not calculated technical indicators from historical OHLCV data.
•FinBERT NLP State: News sentiment classification relies on external NLP microservice calls. Degraded calls display "Sentiment unavailable" rather than false neutral badges.
Full Project Disclaimer

EdgeIQ is a personal portfolio project developed by Raj Puthawala for educational and technical demonstration purposes only. All AI-generated content, scores, and market observations are for informational purposes only and do NOT constitute financial advice, investment recommendations, or an offer to buy or sell securities. Past performance is not indicative of future results. Always consult a qualified licensed financial advisor before making investment decisions.

Future Roadmap

Cloud persistence with Supabase PostgreSQL
Multi-user authentication (NextAuth / Clerk)
WebSocket real-time price streaming
Deterministic OHLCV Technical Indicators
Multi-user journal sharing & export

Raj Puthawala

Full-Stack Developer · AI Integration Specialist

Specializing in Next.js, TypeScript, React, Serverless Architectures & LLM Engineering.

Raj on GitHub