This is for internal testing purposes

● Led a team of 3 to successfully launch LinkedIn's new generative AI job coach feature for members to use natural language queries in job search, for example "Find me software engineering jobs in the bay area". Closely collaborated with XFN partners such as Product and Backend Eng to influence the roadmap and resolve blockers.

● Designed and implemented a unified embedding of member's skills from various sources (explicit, resume, linkedin learning, etc) to consolidate and improve skills representation in downstream models. Improved job recommendation metrics and filed patent for this novel idea.

● Unified 2 LTS AI models recruiter search and recommended matches to reduce maintenance cost and training time by 30%, improving developer productivity and experimental velocity.

Intern on AI foundation team

● Improved salary range predictions on job postings by implementing Monte Carlo Dropout and Deep Ensembles techniques to calculate model uncertainty.

● Implemented and tested the new model end-to-end, including the offline changes to train the model in Tensorflow and deployed to production, resulting in increase in job views with inferred salary.