Working on computer vision R&D for photorealistic avatar generation.

(Codec Avatars)

PyTorch, 3D computer vision

Worked on contents understanding, Newsfeed ranking & recommendation problems to improve user's newsfeed experience. Focused on Facebook Events and Groups content distribution.

* Statistical & machine learning modeling for event / signal prediction.

* Application of various prediction signals to drive user-side value.

* Short & long term forecasting, user behavior modeling, engagement prediction.

Caffe2, PyTorch, Scikit-learn, Pandas, SQL, ranking, machine learning, deep learning, statistical learning

Worked on Iris, a distributed queue with guaranteed in-order delivery powering Messenger and Instagram Direct. Wrote optimizations to payload retention logic for faster end-user experience in Java.

Java, Distributed Systems

Research assistant under Professor Jason Corso, Vision and Perceptual Machines Lab. Special focus on reinforcement/meta-learning and generalizable deep learning methods for navigation problems. Implement various algorithms in TensorFlow and PyTorch.

TensorFlow, PyTorch, deep learning, reinforcement learning, meta-learning

Worked on streaming (realtime) feature engineering pipeline for machine learning projects in Scala and Spark, built a tool to quickly iterate on feature extractors from email data

Scala, Apache Spark, Apache Kafka, Structured Streaming, Data Engineering