Yiren Xu
Building LLM-powered pipelines, distributed systems, and real-time applications. MS Software Engineering @ UC Irvine — GPA 3.93.
Where I've
Built Things
- Built LLM-powered RAG pipeline with document chunking, embeddings, and MongoDB sharding — semantic search at ~200ms retrieval latency
- Designed real-time collaboration system via WebSocket and event-driven state management, enabling AI-assisted editing and reducing frontend latency by 20%
- Built distributed media processing pipeline using Kafka Streams and AWS Lambda with 99.5% task reliability and frontend progress tracking
- Deployed LLM inference services with Ray-style distributed task orchestration and request batching for scalable AI features
- Developed scalable ETL pipeline with Apache Flink and Snowflake — materialized views and partitioning delivered 3× query performance
- Designed NLP service with Gemini API and Redis Cluster caching: <200ms latency for 50+ concurrent requests with token rate limiting
- Implemented full-stack observability with Prometheus and Grafana on Azure VM — 99.9% uptime monitoring
- Built healthcare app with React Native and GraphQL — AI chat interface and WebSocket real-time updates for daily appointment handling
- Designed multi-level caching system (Redis Cluster + local cache) for backend microservices
- Implemented distributed transaction management with local message table and Kafka; high-performance PostgreSQL schemas via table partitioning
- Developed Spring Boot microservices for urban management platform with geospatial data processing — Redis caching and spatial indexing achieving 500 QPS throughput
- Built RESTful APIs with MyBatis, reduced read latency by 30% via custom indexing strategies and cache mechanisms
Technical
Arsenal
Things I've
Created
Peer-to-peer cottage food marketplace with Mapbox-based discovery, real-time messaging, and a compliance-aware AI system. 2025
- Built Supabase schema with Row Level Security and DB triggers; integrated Mapbox for real-time geo-discovery
- Built a compliance-aware AI system using a lightweight RAG pipeline with embedding-based retrieval over food safety policies and allergen data
- Designed an agentic workflow where the LLM orchestrates policy retrieval, ingredient parsing, and compliance validation to generate reasoning-based outputs
- Optimized LLM responses with streaming and batching to improve latency in real-time interactions
IrvineHacks 2024 — inventory management and dynamic recipe suggestions app with React and Spoonacular API integration.
- Check expiration dates and get recipe suggestions from non-expired items
- Responsive UI designed with Figma for cross-device compatibility
- Cut data processing time by 200ms with optimized React hooks
Full-stack mail system enabling users to set email addresses, send and receive messages in real time via REST APIs and AJAX.
- Full send/receive flow with user-defined email addresses
- Real-time updates via AJAX polling
- TypeScript throughout for type-safe API contracts
Academic
Background
Let's
Connect
Currently open to full-time Software Engineer roles. Building at the intersection of distributed systems and AI — reach out if you're working on something interesting.