New Delhi, Delhi, India
Data Scientist with 4+ years of experience building and deploying ML systems in production across forecasting, RAG, and transport management use cases. Specialized in Large Language Models, agentic AI workflows, and reinforcement learning, with a track record of turning research ideas into scalable products on Azure and GCP. Comfortable owning the full lifecycle: problem framing, data engineering (SQL, Snowflake, BigQuery), model design (TensorFlow, PyTorch, XGBoost), MLOps (Kubeflow, KServe, Docker, CI/CD), and monitoring to drive measurable business outcomes. Won 1st place in Blue Yonder’s Crystal Ball AI Innovation event for a reinforcement-learning-based diffusion model that generated optimized truck routes. Recently led development of an agentic RAG chatbot (LangGraph + OpenAI) that enabled the legal team to query 50,000+ contracts and reduced manual review time by 10x. Designed and productionized end-to-end Kubeflow pipelines for training and fine-tuning Hugging Face models, which were later standardized across the data science org. In supply chain and logistics, applied GNNs, RL, and custom quantile regression architectures (LocalGLMNet) to optimize routing and transit-time prediction, improving route efficiency and reducing late deliveries in A/B tests. At IndiaMart, owned one of the largest databases (500M+ records), led migration from Oracle to Postgres, designed ETL pipelines for 250+ database objects, and optimized critical workflows (e.g., user approval process) from nearly 2 hours to a few minutes. Built monitoring and analytics with Grafana to track DB load and scheduled jobs, strengthening the foundation for data-driven applications and ML readiness.