• Collaborating with Prof. Ziang Xiao and Prof. Daniel Khashabi, developed Agentic AI system for LLM HCI research, architecting multi-agent system using LangChain integrated with OpenAI API, implementing prompt engineering and contextual retrieval to structure and improve LLM responses

• Accelerated page rendering speed by 40% through restructuring frontend to component-based architecture with React and hooks, shadcn/ui, Tailwind CSS and managing states by Zustand

• Built Event-Driven backend with Django and Server-Sent Events (SSE), enabling real-time unidirectional streaming from LLM agents to human researchers for interactive evaluation

• Optimized API gateway and boosted throughput by 25% by using Nginx as reverse proxy and load balancer for traffic management and Kafka as message queue for request buffering

• Deployed automated CI/CD pipelines with GitHub Actions to detect build-time errors in Python backend and TypeScript frontend, enhancing system reliability by enabling early issue resolution for scalable AI research deployments

• Course Assistant for EN601.208 Full-stack JavaScript

• Developed a cross-platform sleep tracking and analysis app with React Native and Expo in TypeScript, enabling real-time sleep data visualization and health monitoring

• Designed user-friendly interface with 18+ animated components using React Native Animation API and Reanimated to enhance user experience

• Reduced boilerplate by 60% and improved component re-render performance by migrating from React Context to Zustand for state management

• Engineered Event-Driven microservices using Firebase Pub/Sub and cloud functions for real-time health data processing

• Developed a dynamic product showcase website and trial platform utilizing Next.js, React, FastAPI, and Supabase

• Implemented a high-concurrency e-commerce system for medical consumables ordering, leveraging microservices architecture with Java, Spring Boot, and MySQL, accelerating order processing by 80%

• Developed a responsive user interface using Vue.js and styled components, enabling the sales team to place orders, check inventory, and review order history across web and mobile platforms

• Optimized database performance and reduced data retrieval by 30% by analyzing slow queries with advanced MySQL query optimization and composite indexes creation for large-scale datasets

• Improved consumables lookup time by 96% (from 5s to 200ms)} through implementing two-level cache using Elasticsearch for full-text search, Redis for hot data caching, and LRU eviction strategy during traffic surges

• Ensured reliable order placement during peak bidding seasons by reducing order failures by 99% through implementing RabbitMQ with Topic Exchange pattern to segregate orders from major and minor suppliers

• Reduced release cycles by 80% and ensured consistent production builds via automating CI/CD pipeline through containerized deployments using Docker and Jenkins

• Achieved 100% branch coverage and delivered robust system using JUnit for unit testing and JMeter for load testing