• Designed and developed a cloud-based programming learning platform with Flask and React, boosting user engagement by 40% in just three months by delivering interactive, real-time code execution tailored for beginners.

• Engineered a secure RESTful API with Flask to manage code submissions and leveraged Azure Functions to execute Python code in a protected cloud environment.

• Optimized data retrieval by storing structured data (problem descriptions and IDs) in Azure SQL Server and unstructured data (test case samples and expected outputs) in Azure Blob Storage, which streamlined problem management and sped up code execution.

• Integrated OpenAI's GPT-3.5-turbo API to develop a real-time chatbot that offers problem-solving assistance, hints, and feedback directly in the browser, enhancing user engagement.

Research Assistant at Vanderbilt University Medical Center Health Information Privacy Lab, advised by Professor Bradley Malin.

• Built and normalized over 180,000 EHRs (Electronic health record) training data with given patient information.

• Applied EMR-WGAN on built EHR data to train for best-case models and generated the corresponding synthetic dataset.

• Designed performance analysis metrics including dimension-wise and column-wise distribution, latent cluster analysis, medical concept abundance, clinical knowledge violation and feature importance.

Research Assistant at Vanderbilt University School of Engineering Computational Biology Lab, advised by Professor Maizie Zhou.

• Developed dynamic programming algorithms to identify the SNPs and structure variants in alleles with give BAM files

• Produced Variant Call Format files for 23 pairs of chromosomes and tested on the accuracy of the results by comparing to the reference DNA sequence

• Achieved a 93.8% precision score on detecting SNPs and structure variants

• Contributed to the development of the “Limited Time Promotion” and “Buy Now” features and significantly enhanced sales platform performance, increasing throughput from 500 QPS to 50,000 QPS by implementing Redis and RocketMQ.

• Implemented cache preheating for active promotions and accomplished the lock and inventory reversion using Redis and Lua script.

• Managed peak loads on the order service by asynchronously creating orders through RocketMQ and implemented the "Pay-Check & Cancel" feature by consuming delayed messages.

• Optimized the optimistic-lock approach when updating the inventory in the MySQL database with Repeatable Read isolation level.

• Applied the principles of Domain Driving Design (DDD) to separate the order service into promotion, order, and payment processing microservices.