Snapchat messaging team

• Engineered core components of a release management platform, enhancing automation for mobile app release decisions impacting 600M+ users.

• Worked on high-throughput, multithreaded data pipelines to process ~50k stability metrics weekly, optimizing data ingestion from Crashlytics and internal analytics and data store.

• Developed a centralized release analytics dashboard, streamlining changelogs and crash data for efficient beta and production release cycles.

• Developed and deployed critical services for the 'Reclaim' product's billing SDK, enhancing user subscription management and payment processing.

• Revamped the password manager SDK, focusing on vault enhancements and significantly reducing unlock times.

• Collaborated with cross-functional teams to ensure seamless integration and deployment of new features, improving overall user experience.

• Enhanced the Colgate Connect Android app by implementing automatic Bluetooth synchronization with smart toothbrushes, improving user interaction.

• Researched deep learning models to analyze brushing techniques using IMU sensor data, providing actionable insights on effectiveness.

• Developed a script for an internal brushing study to gather data for model training, contributing to product innovation.

• Developed a SpeechToText model for Boeing Inc, combining 5000 hours of unbiased dataset and generating 10% more data using selected YouTube videos for enhanced training

• Optimized Mozilla’s deep speech model with CTC Beam Decoder for noisy airplane environments, achieving a 20% increase in recognition performance compared to the unmodified model

• Improved speech recognition accuracy in noisy airplane environments (60% SNR) by 15%, utilizing additional noises in the dataset, applying DeepDenoiser for noise reduction, and conducting rigorous testing