a software engineer.">
Jan 2019 - Apr 2019
Designed, developed, and deployed a presubmit analyzer on Android Gerrit that computes and displays API coverage information for each API addition/modification included in a Gerrit Change, using Golang
Open source contribution to the CTS API Coverage Analysis tool, using Java
May 2018 - Aug 2018
Developed the full stack of an internal tool for managing test instances; backend using Golang, AWS API Gateway, Lambda, and RDS (MySQL); frontend using React.js, a serverless webapp built on AWS S3; deployed using Docker
Researched and implemented Fast Robust PCA model (paper) for anomaly detection; implemented with Scala
Feb 2018 - May 2018
"Deep Learning from Logged Bandit Feedback"
Researched a counterfactual risk minimization on deep learning using variance regularization; demonstrated how neural networks can be effectively trained for object recognition and banner-ad candidate selection without fully labeled data
Sep 2017 – Dec 2017
"Active Learning for Multi-Label Classification"
Researched a way to reduce selection bias in an active learning using stochastic sampling, and to normalize the loss using inverse propensity; experimented with TensorFlow
Jun 2017 – Aug 2017
Built a data pipeline and API using Boto3, SqlAlchemy, and Flask to store and retrieve data from AWS
Developed a multi-label multi-layer perceptron classifier that predicts the tags and genres associated with each unknown song given; used Scikit-Learn and TensorFlow for implementation, Python Librosa for analysis