AI Engineer focusing on Multi-Agent Systems and LLM Fine-Tuning. I engineer functional AI systems that bridge the gap between AI research and practical applications.

• 5+ AI Projects engineered using LangGraph, FastAPI, and Gemini 2.0.
• Specializations: Agentic Workflows · RAG Pipelines · MCP Servers · Computer Vision.
• Key Projects: Engineered modular DPO alignment pipelines and autonomous agent swarms.
• Location: Navi Mumbai, Maharashtra, India.

QuantIQ is a full-stack decision intelligence platform that integrates real-time market data, technical indicators, machine learning, and AI agent reasoning into a unified trading terminal.

• Real-Time Data Flow: Ingests live stock price ticks from Yahoo Finance every 5 seconds, routed via a Redpanda (Kafka-compatible) message broker to browser clients using WebSockets.
• Local ML Inference: Engineered an on-device ONNX model (RandomForest) inside the FastAPI process to predict stock price direction with <5ms latency and zero API cost.
• Multi-Step AI Analyst: Built a Gemini 2.5 Flash ReAct agent that executes custom Python tools to fetch stock history, compute indicators (RSI, MACD, EMA), and run ML predictions to generate structured financial reports.
• MLOps Feedback Loop: Designed a Celery background worker that automatically labels and evaluates past predictions, comparing user-defined trading strategies against AI-recommended targets over time.
• Observability: Integrated custom Prometheus metrics to track token usage, agent latency, and data pipeline delay, visualized on a Grafana Cloud dashboard.
• Tech Stack: React 19, Vite, TypeScript, FastAPI, Redpanda (Kafka), NeonDB (PostgreSQL), Celery, Redis, ONNX Runtime, Gemini 2.5 Flash, Razorpay.

Metric	Value
Total Repositories	46 Public
Total Stars	⭐ 75
Total Forks	🍴 8
Followers	👥 11
Following	👤 4
Last Updated	November 12, 2026 at 06:28 PM UTC

1. ❌ Merged PR #21 in Edge-Explorer/Tracepass