Transfer Learning for Crisis Response
Used RoBERTa to transfer knowledge across organizations to improve classification effectiveness for organizations with smaller amount of available training data. Work done as part of the AMLD 2020 Challenge.
AppService for Building Depot
Building a managed AppService using Flask and Nginx for hosting IoT apps in discrete LXC containers. Created REST APIs for other systems to be able to interact with AppService and deploy third party apps. Work done as part of Synergy Labs, CMU.
Workplace Automation and Server Management
TARS is a bot that can auto-verify assignments, track student progress, provide updates on office hours, run polls and more in a slack workspace. Work done as part of Solarillion Foundation.
Pollen Grain Classification
Built a two-stage segmentation and classification approach to classify pollen grain images. Work will be presented at ICPR 2020 Challenge Workshop.
Forecasting the On-time performance of flights
Designed a two stage predictive Machine Learning Engine that forecasts the On-time performance of flights using Weather Data in the US.
Spotify Sequential Skip Prediction
Predicted whether individual tracks encountered in a listening session will be skipped based on the first half of a user’s listening session. Submission ranked in the top 3% in the challenge jointly conducted by Spotify and AICrowd.