As part of Worldwide Sustainability team, currently working on LLMs for Sustainability Applications

1. Developed a deep learning-based generative framework to extract conditional feature representation, and generate conditional synthetic data to tackle data insufficiency challenges.2. Improved the generalizability of personal thermal comfort prediction algorithms by 16% via transfer learning. 3. Developed deep learning algorithms for occupant activity detection in smart buildings using privacy-preserving WiFi signals. Improved the performance by including camera signals wherever ethically possible via sensor fusion. The system achieved a 97.5% accuracy in occupant activity detection.4. Designed a graphical lasso-based framework to intelligently segment the players in energy game-theoretic frameworks into energy usage groups, crucial in smart incentive design.

1. Conducted an extensive survey on a new area of research- Foundation Models (FM) and its implementation. 2. Proposed a research approach to using FMs in a distributed learning setting, combining training, aggregation, and online temporal updating of FMs. 3. Designed a proof-of-concept implementation of above approach to introduce high performance and communication-efficient federated multi-task learning for ByteDance products.

As part of Amazon Alexa AI,1. Worked on the NLP engine of Alexa to enhance its contextual question-answer capability. 2. Implemented Contextual Bandits and Prioritized Experience Replay to improve follow-up question-answer performance by 17%. 3. Introduced and implemented an Online Learning pipeline for continual training and adaptation of the model.

1. Developed sink interface of Embedded Display Port (eDP) protocol for energy-efficient single-stream transport using System Verilog and C++. 2. Improved the simulation verification mechanism by developing a universal check script in Perl.

- Formulated and analyzed the Business Case of a weight saving proposal with 183.7 kg saving per A350-1000 aircraft.- Nominated by team to support live projects on A350-900 and A350-1000 Electrical Model Design.