Building production-grade AI systems that solve real-world problems. Specialized in RAG pipelines, LLM fine-tuning, prompt engineering, and scalable ML deployment. From concept to deployment, I create AI solutions with measurable impact.

🏆 Best Outgoing Student 2025 | 🎯 TNSCST Grant Winner (₹20K) | 🥇 Hackathon Champion

• Generative AI & RAG Systems: Building production RAG pipelines with 82% hallucination reduction
• LLM Fine-tuning & Optimization: PEFT, prompt engineering, context optimization
• ML & Deep Learning: End-to-end pipelines from data engineering to deployment
• AI Architecture: Vector databases, graph DBs, state machines, and LLM orchestration
• MLOps: Docker, FastAPI, model versioning, evaluation frameworks

Machine Learning & Deep Learning
Deep Learning (PyTorch) • Scikit-learn • Mathematics & First Principles • ML Algorithms

Generative AI & NLP
LLM Fine-tuning (PEFT) • Transformers • Vector Databases (ChromaDB) • Graph Databases (Neo4j) • Embeddings • RAG Pipelines • Tokenization • Prompt Engineering • Context Optimization

MLOps & Deployment
Model Packaging & Versioning • Docker • Git/GitHub • REST APIs • FastAPI • Flask

Tools & Frameworks
LangChain • Hugging Face • Apache Airflow (basics) • Google Cloud Platform

Production-grade RAG system with LLM orchestration and multi-model ML pipeline

• 🎯 82% hallucination reduction using Delta State Graph architecture
• 📊 5 predictive ML models (churn, demand forecasting, sentiment, anomaly, cohort analysis)
• 🔄 RAG pipeline with ChromaDB vector database + SQL cross-validation
• 🛠️ 65% reliability through context optimization and prompt engineering
• 💻 Tech: Python, PyTorch, LangChain, ChromaDB, FastAPI, XGBoost
• 🧬 PEFT (LoRA) for efficient fine-tuning on limited compute
• 📊 Custom dataset curation from 42K+ real-world samples

Key Innovation: Designed a state-tracking system that separates known/missing/assumed data to minimize LLM hallucinations and improve answer reliability.

Graph-based RAG system for supply chain disruption prediction

• 📈 200K+ nodes modeled in Neo4j for multi-hop disruption tracing
• 💰 5-10 crore revenue loss prevention through predictive stockout scenarios
• ⚙️ 45% token cost reduction via custom DCBA algorithm (cuts redundant LLM calls)
• 🔗 Knowledge Graph + RAG for intelligent supply chain insights
• 💻 Tech: LangChain, Neo4j, OpenAI, Graph Neural Networks, Custom Algorithms

Business Impact: Prevents critical stockouts for high-velocity e-commerce by predicting disruptions 3-5 days in advance.

Multi-feature AI platform for adaptive learning and engagement

• 🤖 AI chatbot with voice/text support (Gemini API + Speech-to-Text)
• 🎮 Object recognition games using YOLO and TensorFlow
• 📚 Adaptive content tailored for different age groups
• 📊 Engagement tracking with real-time analytics dashboard
• 💻 Tech: React, MongoDB, TensorFlow, YOLO, Gemini API, Speech Recognition

Research: Published LumiLearn paper on personalized learning using ML-based content adaptation and GenAI recommendations.

AI/ML Specializations:

• IBM Python for Data Science

Development & Cloud:

• Python Full Stack Developer – IPCS Global
• Google Cloud Essential Training – Pearson & LinkedIn
• Salesforce Platform Developer – 13 Superbadges, 28 Badges

I'm currently seeking AI/ML Engineer roles where I can build production-grade AI systems that create real impact.

"I don't just build AI models—I architect production systems that solve real business problems. From reducing hallucinations by 82% to preventing crore-scale revenue losses, my work delivers measurable impact. If you're building AI products that matter, let's talk."

Open to: Full-time AI/ML roles | RAG/LLM projects | Generative AI consulting

Location: Coimbatore, Tamil Nadu, India
Phone: +91 6385225238

⭐ If you find my work interesting, consider starring my repositories!