AI/ML Engineer | Building production systems that scale | Training models at 3 AM
Currently pursuing B.Tech in Data Science & AI at IIT Bhilai
I build AI systems that work in production, not just in notebooks. My focus is on LLMs, multi-agent architectures, and making ML models actually deployable at scale.
- Currently Building: Human Slop — Anti-AI social platform with hardware-bound authentication
- Published: pytorch-dml — Deep Mutual Learning library on PyPI
- Writing: 70+ ML/AI articles on Medium with 800+ monthly readers
- Recent Won: Silver Medal on Kaggle (Rank 36/1,711) in MITSUI Commodity Prediction Challenge
Production-ready PyTorch library for Deep Mutual Learning and collaborative neural network training.
- Impact: 2–5% accuracy improvement over independent training
- Scale: 7,000+ lines of code, 34 modular components
- Features: AMP, DDP, ONNX export, full test coverage
- Stack: PyTorch, CUDA, Distributed Training
Human Slop — Anti-AI Social Platform
Privacy-first social platform that blocks 100% of AI-generated content using typing forensics.
- Architecture: Dual-database system, hardware-bound biometric auth
- Innovation: Real-time keystroke analysis (WPM, burst patterns, pauses) → Human Score
- Stack: FastAPI, React Native, Supabase, Device Fingerprinting
- Status: Fully deployed (Web + Mobile)
Hedgera — Autonomous Financial Intelligence Platform
Multi-agent AI system for autonomous market analysis and portfolio management.
- Performance: ~20% returns with 5–8% max drawdown (backtested + paper traded)
- Architecture: 7-layer system with temporal data fabric, RL forecasting, agentic debate framework
- Team: Led 8-member engineering team
- Stack: Pathway, PyTorch, LangChain, Real-time streaming
Kerala Ayurveda -A Medical RAG System
Production-grade multi-agent medical information retrieval system.
- Accuracy: <10% hallucination rate (RAGAS), 35% retrieval improvement
- Architecture: CRAG framework with hybrid BM25 + vector retrieval
- Stack: GPT-4, LangGraph, ChromaDB, FastAPI, Streamlit
Team Lead — Autonomous Financial Intelligence Platform
Independent Research Project, IIT Bhilai | Oct 2025 – Dec 2025
Led 8-member team building multi-agent financial AI system combining real-time streaming, RL, and LLM reasoning.
AI Developer Intern — Kartavya Technology
Remote | Jun 2024 – Aug 2024
Built multi-agent automation systems and cloud APIs on AWS/GCP. Reduced manual effort by 40%, maintained 99.9% uptime, cut infrastructure costs by 25%.
Coordinator — Data Science & AI Club, IIT Bhilai
Led workshop,and hackathons promoting ML culture within the college.
Languages: Python, C++, SQL, JavaScript, Bash
ML/AI: PyTorch, TensorFlow, Scikit-learn, Hugging Face, XGBoost, OpenCV
LLMs & Agentic AI: GPT, LangChain, LangGraph, RAG, Fine-tuning, Prompt Engineering, Multi-Agent Systems
MLOps: MLflow, DVC, SHAP, Model Versioning, Experiment Tracking, Docker, Kubernetes
Infrastructure: GCP, AWS, CUDA, FastAPI, REST APIs, CI/CD, Distributed Training
Data: PostgreSQL, MongoDB, Neo4j, FAISS, ChromaDB, Pinecone
- Kaggle Silver Medal — MITSUI & CO. Commodity Prediction Challenge (Rank 36/1,711)
- Kaggle Top 10% — GQ Volatility Forecasting (Rank 34/386)
- Amazon ML Challenge 2025 — All-India Rank 278
- Winner — Pixel Perfect Hackathon (IIT Bhilai) — Improved baseline by 23%
- Top 10% — Data Science Bootcamp (IIT Guwahati) — Ranked among 500+ participants
- Technical Writer — 70+ articles on Medium, 800+ monthly readers
- YC Startup School India — Bengaluru 2026
"Reducing loss functions in both code and real-world problems."
If you find my work valuable, consider starring my repositories!