Building large-scale LLM serving and inference systems for Tinder

Automated the verification of over 4.5 million ads every day, enhancing machine-learned models’ performances.

Adopted serverless functions with event-driven DynamoDB Streams to process data and store results in S3 Bucket.

Employed Typescript CDK to manage the AWS cloud infrastructure while ensuring scalability and fault tolerance.

Researched various technologies and presented the solution to Amazon’s Ads teams in an org-wide Tech Talk.

Achieved over 90% unit and integration test coverage for six modules adopting Karma and Jasmine test framework.

Spearheaded the design and development of a full-stack Timesheet application using Vue.js, PostgreSQL, and Node.js to automate employee time sheets for thousands of users, saving over 10 person-hours every week.

Implemented RESTful APIs leveraging Node.js framework while choosing the right HTTP methods, headers, and response codes to ensure idempotency and graceful API retries.

Awarded Move The Dot and Live The Dot for outstanding work commitment and leadership in various projects.

Worked with SQLite database to manage user data and Android Studio utilizing Java as a development environment.