Jan 2023 – Present · 3 yrs 4 mos

Jul 2021 – Mar 2023 · 1 yr 8 mos

Feb 2020 – Jul 2021 · 1 yr 5 mos

• Build unified identity authentication for Confluent Cloud ecosystem

Oct 2019 – Dec 2019 · 2 mos · Greater San Diego Area

Jun 2019 – Sep 2019 · 3 mos · San Francisco Bay Area

• Software Engineer Intern in Release Engineering Team;
• We own the distributed test infrastructure that ensures that changes are safe before getting rolled out to users;Release engineers work on one of the largest Mesos clusters at Yelp, collaborate on their in-house test results platform, and build the next generation of developer metrics.
• Projects:
• ◦ Built a real-time short-URL search tool during Yelp Hackathon that is used by over 500 Yelp engineers and gives possible URL suggestion when encountering 404 error.
• ◦ Refactored RESTful APIs by pagination and lazy loading to reduce slowest web access from 27s to 0.5s.
• ◦ Integrated cProfile into Kafka consumers to collect profiling data and detect performance bottleneck.
• ◦ Improved the message throughput of Yelp Engineering Effectiveness tool’s data pipeline from 3000
• msgs/minutes to 33900 msgs/minute by Python LRU Cache, load balancing and Batch SQL.

Aug 2017 – Jun 2018 · 10 mos · San Francisco Bay Area

• ◦ One year full-time internship in CTAO healthcare team which built Smart Connection Solutions aimed at helping providers connect with patients before, during and after care delivery for Optum.
• ◦ Distributed Meraki Data Analysis Platform: Implemented and deployed a distributed real-time log analysis system to monitor and predict Cisco Meraki cloud platform’s status based on HSMM(Hidden Semi-Markov Model) and Elastic Stack, which can reach a high predictive accuracy of 77.3%.
• ◦ Huddle Dashboard: Created a dashboard based on Kibana and Bootstrap, to integrate and display patients’ health-related data from three different sources.
• ◦ Huddle API: Deployed scalable and elastic REST API microservice with Spring Boot and Kubernetes as the backend for Cisco Health Care Huddle system, to support online conference between patients and medical teams.
• ◦ Voice-controlled Assistant: Added a voice-controlled module into the Tumor Board application. Combined Redux and ReactJS with State Machine theory to control UI’s refreshing. Utilized TensorFlow to build a CNN model and improved the accuracy rate of intention classification from 82.7% to 97.8%.