Resume

Software engineering internships, projects, and research in LLM backend systems. Seeking Summer 2027 SWE internships.

Download PDF

Education

University of Southern California, M.S. in Computer Science, Jan 2026 - Dec 2027
GPA: 4.0/4.0 · Coursework: Analysis of Algorithms, Machine Learning for Data Science

Southern University of Science and Technology, B.S. in Computer Science, Sep 2021 - Jun 2025
GPA: 3.60/4.00

Experience

Figwork | Software Engineering Intern (Part-time, Remote)

Sep 2026 - Present | Sunnyvale, CA

  • Developed LLM and backend systems for an AI recruiting platform (NestJS/TypeScript, PostgreSQL, AWS Bedrock), building the agent that turns a recruiter's chat into a structured hiring spec and the backend services behind candidate matching.
  • Co-architected a two-stage candidate recommendation system over 160K+ profiles (embedding recall + LLM reranking against a 106-field job schema) that learns each recruiter's preferences online from like/skip feedback with cold-start exploration.
  • Migrated candidate retrieval to pgvector and accelerated hiring-spec insights with caching and an incrementally refreshed inverted index, cutting end-to-end insights latency from 22s to 0.2s.

Segi | Software Engineering Intern, AI Platform Development

Jun 2025 - Dec 2025 | Shenzhen, China

  • Developed core backend services for AI customer support and workflow automation on a property-management platform serving 3 enterprise clients across 60+ properties; workflows processed 11,800+ tickets and cut ticket creation time from 2 minutes to under 30 seconds.
  • Engineered multi-tenant WhatsApp/WeCom message ingestion in Spring Boot with incremental polling, Redis per-tenant locks, TTL deduplication, and DB-backed cursor recovery to prevent duplicate or lost messages.
  • Designed a modular agent workflow engine (Spring Boot, LangChain4j) with reusable nodes and conditional routing for intent recognition, parallel multi-query RAG retrieval with Milvus, and MCP tool execution.
  • Built fault-tolerant AI request handling with bounded thread pools, timeouts, retries, circuit breaking, and fallback replies, using RabbitMQ to absorb traffic spikes and raising AI reply success from 83% to 96% in 200+ concurrent load tests.

MGI Tech | Software Engineering Intern, Algorithm Development

Jan 2023 - May 2023 | Shenzhen, China

  • Built perception algorithms for a robotic ultrasound system: fine-tuned AlphaPose on supine patients to localize the xiphoid process for robot-arm positioning, and cut image origin-detection latency from 200 ms to 11 ms.

Projects

NextTier: AI-Powered Custom Merchandise Platform | Industry Project Lead

Apr 2026 - Jul 2026 | Los Angeles, CA

  • Built and deployed an AI custom merchandise platform on AWS with React and FastAPI microservices for design generation, product previews, and ordering; validated through 500 simulated end-to-end workflows.
  • Designed PostgreSQL schemas for products and carts and added Redis caching, reducing database reads by 69% and improving API latency by 32%.
  • Integrated AWS Bedrock into backend generation services with request validation, duplicate-request protection, and invocation safeguards.
  • Implemented asynchronous order processing on Amazon SQS, reducing P95 checkout latency by 77% at 100 concurrent users with under 1% failures.

Research

CUA-SWE: When Computer-Use Agents Meet Visual Software Engineering
Co-first Author · FORTIS Lab, USC · Jun 2026 - Sep 2026 · Under review at ICLR 2027

  • Curated 33 web repair tasks through an adversarial curation loop from real GitHub issues and observed UI failures.
  • Built deterministic behavioral verifiers and analyzed paired code-only and CUA trajectories to diagnose partial repairs, protocol violations, and frontier failures.

Publications

  • Li, Z., Ling, Y., Gu, Y., Huang, B., Liang, C., Islam, S., … & Seita, D. FELT: Generating Tactile Signals from Vision for Visuo-Tactile Manipulation. Tactile Sensing for Robotic Foundation Models Workshop, RSS 2026.
  • Lin, H., Li, H., Li, H., Yu, X., Yu, K., Liang, C., … & Liu, J. Data Heterogeneity-Aware Personalized Federated Learning for Diagnosis. OMIA Workshop, MICCAI 2024.

Technical Skills

Languages: Java, Python, JavaScript, TypeScript, C/C++, SQL, HTML, CSS
Frameworks: Spring Boot, FastAPI, NestJS, React, Next.js, LangChain, LangChain4j, Prisma
Databases: MySQL, Redis, PostgreSQL, Milvus, MongoDB
Tools: AWS, Docker, Kubernetes, Jenkins, Git, Linux, Maven, Nginx, RabbitMQ