AI/ML Engineer building production-grade systems that solve real-world problems
I specialize in taking complex technical challenges and turning them into deployed solutions that people actually use. From HIPAA-compliant healthcare AI to legal document automation, I've built systems across domains that require both technical depth and practical execution.
📍 Location: Ghaziabad, India
💼 Current: AI Engineer @ LetsAI Solutions
✍️ Writer: 12,000+ views on Medium covering production AI systems
Built HIPAA-compliant medical AI system using Graph RAG architecture with automated PII anonymization. Integrated BioBERT for medical entity recognition and intelligent patient-doctor matching based on symptom classification.
Tech: Graph RAG, BioBERT, NER, HIPAA compliance, Multi-document processing
Engineered system generating 150+ page immigration petitions in 45 minutes (vs. 50+ hours manually). Fine-tuned Llama 3.1 8B on 6,500+ curated examples using QLoRA, achieving 91% accuracy on gap detection vs. expert attorneys.
Tech: LangGraph, QLoRA fine-tuning, Gemini 2.5 (data curation), ChromaDB, Long-context generation
First production implementation of Azure SQL vector search in India. Built multimodal RAG system integrating images and videos into AI tutoring sessions, achieving 99.7% uptime for 10,000+ students.
Tech: Azure SQL Vector (pioneer), Multimodal RAG, Smart media indexing, Production deployment
expertise = {
"advanced_architectures": ["Graph RAG", "Multimodal RAG", "Long-context generation (150+ pages)"],
"fine_tuning": ["QLoRA", "Dataset curation (6,500+ examples)", "Model evaluation"],
"production_systems": ["99.7% uptime", "10K+ users", "HIPAA compliance"],
"specialized_domains": ["Healthcare (BioBERT)", "Legal AI", "Education"],
"multi_agent": ["LangGraph", "CrewAI", "Complex orchestration"],
"vector_dbs": ["Azure SQL Vector", "ChromaDB", "Pinecone", "MongoDB Atlas"],
"llms": ["GPT-4o/o1", "Claude", "Llama 3.1", "Gemini 2.5"]
}
I write about production AI systems, not just tutorials. 10,000+ readers on Medium
- "The Evolution of LLM Inference: Arctic Inference Revolution" (Feb 2025)
- "Fine Tuning LLM Guide: From Dataset Curation to Production" (Feb 2025)
- "Azure SQL for Vector Databases: Revolutionizing RAG" (Nov 2024)
- "Building Image Vector Stores for Multimodal RAG" (Oct 2024)
Fine-tuned model on 6,500+ denied petitions achieving 91% accuracy in identifying gaps that lead to RFEs. Automated data curation pipeline using Gemini 2.5.
Impact: 98% time reduction in petition review (50 hrs → 45 min)
Graph RAG system with automated PII anonymization for healthcare deployment. Uses BioBERT for medical entity recognition.
Impact: HIPAA-certified deployment for hospital chain
Azure SQL vector search implementation integrating images and videos into tutoring sessions. First production deployment in India.
Impact: 10,000+ students, 99.7% uptime
Multi-class sentiment classifier achieving 91% accuracy. Deployed on HuggingFace Hub.
B.Tech in Information Technology | JSS Academy of Technical Education (2019-2023)
Career Trajectory:
- 2025-Present: AI Engineer @ LetsAI Solutions - Legal document automation, fine-tuning
- 2024-2025: AI/ML Engineer @ Techoon Solutions - Healthcare AI, educational platforms
- 2023: Research Analyst @ Quaintel Research - Market intelligence, data pipelines
✅ Building complete systems, not just prototypes
✅ Production deployments with real users (10K+)
✅ Advanced RAG architectures (Graph, Multimodal, Long-context)
✅ Fine-tuning expertise with custom dataset curation
✅ Working in regulated domains (HIPAA, legal)
✅ Technical writing that explains complex systems clearly
I'm always interested in discussing:
- Production AI systems and architecture decisions
- Fine-tuning strategies and dataset curation
- RAG optimization (especially Graph RAG and multimodal)
- Deploying AI in regulated domains (healthcare, legal)
- MLOps and maintaining 99.7% uptime
⭐ If you find my work valuable, star the repositories you find interesting!