Applied Scientist Intern at Amazon Web Services (AWS), Palo Alto, CA
Mentor: Anima Anandkumar, AWS AI team
Project: Action Recognition on Compressed Videos
Description: Designed and evaluated algorithms to perform video encoding based action recognition using a combination of 3D CNNs and LSTMs on compressed videos for UCF101 and Charades dataset.
Applied Scientist Intern at Amazon, Austin, TX
Mentor: Ali Jalali, Mobile Homepage Personalization Team
Project: Personalized recommendations using passive-aggressive algorithms
Description: Improved the ranking of products on the mobile Amazon homepage for users by implementing passive-aggressive algorithms and a random forest ensemble learning to achieve improvements of 3 % and 5 % respectively.
Research Intern at Technicolor Research Bay Area, Los Altos, CA
Mentors: Nikhil Rao, Weicong Ding, Machine Learning Team
Project: Higher Order Matrix Factorization for Recommender Systems
Description: Designed an algorithm to make use of implicit higher order information for predicting unobserved entries in a ratings matrix. Achieved improvements for MovieLens, FilmTrust and Gene datasets using the NDCG and MAP metrics.
Research Intern at KTH, Stockholm, Sweden
Mentor: Alexandre Proutiere
Project: Minimizing System Backlog in Wireless Networks using Restless Bandits
Description: Modeled the downlink scheduling in wireless networks as a Restless Bandit Problem. Derived the existence of Whittles index criteria for the problem of minimizing system backlog with delayed queue state information.