Reduced build times of core Python services by 90% by improving caching

Collaborated to create a new cloud live streaming system using websockets, fragmented MP4, Media Source API, and concurrent Go code. System improved live video latency from 4s (SD) and 12s (HD) to 1-2s.

Architected OAuth2 flow and endpoints for internal support tools’ new web interface using Flask, Auth0, and Okta. Lead to adoption by 130+ monthly users and positive feedback.

Created watchdog system to monitor for outages using live endpoint and browser synthetic tests. Alarms it triggered have been the first to go off during multiple outages. System has been adopted company-wide.

Developing system to test deep learning action recognition systems using synthetic data from video games

Designed and implemented video annotation tool for Amazon Mechanical Turk

Organizing TA team to hold office hours

Helping students in office hours and online Q/A wiki

Evaluating students’ papers analysing performance of machine learning and optimization algorithms

Implemented API endpoints and front end to configure AWS provisioning software

Empowered support engineers to quickly make changes to resolve internal issues

Used Java, Spring Boot, Hibernate, Angular, Jenkins, and JAX-RS

Enabled users to make predictions with ARIMA regression models and recurrent neural networks

Used Python, Tensorflow, PyQt, Numpy, Scikit Learn, Pandas, and Statsmodels