At HCLTech, I implemented a T-LSTM environment model that reached 97% accuracy on a client-cleaned dataset, and compared it with BiLSTM and SSM architectures. I also built DQN, TD3, and PPO agents to test reward design and environment reliability on temporal medical data.
For HCLTech’s agriculture capstone, I built an agentic system that brings together multiple viewpoints and a final summary. I set up a PostgreSQL database for the real-world ICRISAT dataset and implemented an NL2SQL tool and a PostgreSQL/pgvector RAG pipeline.
At Amazatic, I owned and led development of three end-to-end Redshift and DBT data pipelines for sales and advertising revenue decisions, with corresponding dashboards. I also developed an insights dashboard with product-level metrics, time-series analysis, and XGBoost predictions.
Earlier, at Microsoft (Contract), I created Power BI and Kusto cluster health reports and enhanced a feature across .NET Core microservices. At AWM Smartshelf, I developed computer vision algorithms for spatial awareness and deployed them for live clients.

