Working on multi-cloud model training + serving infrastructure and cloud developer environments to power diversified ML/AI use cases across several verticals like finance, healthcare, logistics and many more.

Teaching Assistant for CIS - 520: Machine Learning for Fall 2021, Spring 2022. CIS-520 is a graduate level course and one of the largest classes offered at Penn, with a cohort of 100-200 undergrad/master's/PhD students.

Worked under Dr Osbert Bastani on adversarial attacks against reinforcement learning enabled software systems.

Research assistant at Kumar Research Lab (GRASP). Working on the DCIST project which aims to build autonomous, resilient and heterogeneous swarms of robots that can augment or replace human effort in a variety of scenarios. More @ https://www.dcist.org/

Worked in Microsoft Search on powering acronym answer on bing.com, sharepoint.com and office.com endpoints. My work spanned across serving and extraction/indexing infrastructure. The goals here were to build and maintain a highly available and low latency search infrastructure which guarantees data consistency, avoids any data loss and is resilient towards failures. Across all endpoints my work was part of serving typically 500M - 1.5B requests a day.

Previously as part of the Microsoft Assistant (Cortana) Notifications team, I was involved in building a decentralized notification platform from scratch, that could serve millions of users. The goals of the new platform were high availability, better fault tolerance, improved developer agility and decoupled deployments.