Current responsibilities--Enhance Large Language Model inference efficiency through testing and optimizations.--Ensure data quality and reliability for accurate model outputs.

-- Worked as a part of the Alexa ASR team within the Amazon AGI org, tasked with delivering comprehensive speech recognition solutions to both internal and external customers.-- Developed, tested, and deployed On-Device Speech2Intent (S2I) ASR model for a prominent luxury automotive manufacturer, achieving a 35% reduction in word error rate and a 25% improvement in intent recognition error rate for the S2I model.-- Created a mechanism enabling both Alexa ASR models and S2I models to support a new wakeword, reducing false rejects from 75% to under 5%.-- Designed and implemented a robust data pipeline for processing third-party (3P) data, enhancing evaluation benchmarks, and replicating the production environment.-- Acknowledged for resolving a critical issue that was hindering the deployment of a substantial model for a luxury automotive manufacturer.-- Actively engaged in and initiated multiple critical code reviews, contributing to elevated code quality.