CS6515: Intro to Graduate Algorithms

• Design system for building embeddings from personalized contextual signals to be used for context retrieval and injection.

• Implemented algorithms and APIs on measuring phonetic similarity as part of a system for contextual recovery of out-of-lattice named entities in speech recognition.• Developed mechanism for generating and serving phonetic contextual models from known entities asynchronously on-device.

• Led design and implementation on a system for bringing contextual information to speech recognition queries from biasing grammars across different languages.• Improved quality of recognition for MVP queries issued when launching Assistant in new languages.

• Underwent three-month training period consisting of group projects, workshops, and lectures learning Google tools and infrastructure as well as general industry best practices.• Researched the use of MultiTask Learning to unify acoustic models for the same language across dialects for first rotation with a Speech team.• Implemented pipeline using MapReduce framework for measuring precision and recall for inferred location from logs data for second rotation with a Google Maps team.

CS7637: Knowledge-Based Artificial Intelligence

• Automated the task of finding and indexing all the company’s research and publications.

CS102: Data Structures