Building reliable AI systems that move from research to real-world impact.
I design and build production-grade LLM applications, retrieval systems, and agentic workflows for regulated and high-scale environments across BFSI, healthcare, and climate-tech.
Over 8+ years I've moved from classical ML and forecasting to leading GenAI architecture at enterprise scale — shipping systems for BFSI, healthcare, environmental intelligence, and people analytics, always with an eye on what actually survives contact with production.
Featured Projects
Agentic Customer Support AI
End-to-end multi-stage reasoning system for support resolution — retrieval, contextual generation, and an escalation path for sensitive cases.
Enterprise Multi-Agent Platform
Architected multi-agent AI systems for a global financial institution, automating manual controls and compliance workflows — from document understanding to validated, auditable outputs.
LoRA Fine-Tuning + RAG
Parameter-efficient fine-tuning of GPT-2 with retrieval augmentation layered in — full performance, a fraction of the compute cost.
Areas of Expertise
Enterprise GenAI
Designing and deploying scalable, secure GenAI solutions in regulated environments.
Retrieval Engineering
Hybrid search, reranking, and RAG pipelines, evaluated with Precision@K, Recall@K, and MRR.
Agentic AI
Planning, memory, tool use, and multi-agent orchestration for complex workflows.
ML Engineering
Model training, fine-tuning, deployment, monitoring, and optimization at scale.
AI Governance
Evaluation frameworks, guardrails, and human-in-the-loop compliance design.
Latest Insights
Experience
Built a multi-agent platform that turns unstructured enterprise documents into structured knowledge, plus a GenAI validation framework for governance and compliance.
Built NLP intent-recognition with Clinical BERT, lifting lead-to-MQL conversion 20%, and Bayesian market-mix models for channel investment decisions.
Built cloud-native geospatial and time-series forecasting pipelines on AWS (XGBoost, ARIMA, LSTM) for air-quality prediction at scale.
Built automated ML pipelines for workforce analytics, cutting model retraining time 90%.
Let's build something worth shipping.
Building at the intersection of AI, engineering, and research. Always open to meaningful conversations and collaborations.