Photo of Trung Duong

CS + Linguistics @ UCLA · SWE Intern @ Serafis, YC S25

I build full-stack products that make complex data usable.

UCLA CS + Linguistics student and SWE intern at Serafis (YC S25), focused on React/TypeScript, API-driven products, data visualization, and ML-backed tools.

  • 8+ product surfaces shipped
  • 10+ REST workflows integrated
  • 100+ UCLA pathways supported
  • 30% faster ML training
View ProjectsDownload ResumeContact Me

Featured Work

Shipped product engineering, full-stack tools, and ML research with measurable impact

Production

Serafis

Production React/TypeScript app — 0-to-1 launch as 1 of 3 core contributors · YC S25

  • Shipped 8+ product surfaces: Dashboard, Podcasts, Episodes, Companies, 4 AI Search tabs, and Ask AI
  • Rebuilt search with hit-count relevance — 12% faster discovery, 8% more selections, 15% less feedback
  • Integrated 10+ REST workflows including 3-tier billing, watchlists, PostHog, and AI citations
ReactTypeScriptViteTailwindTanStack QueryFirebase
ExperienceResume ↗

Project

TreeReq

Deployed prerequisite-tree visualizer for 10+ UCLA students across 100+ majors/minors

  • Interactive D3.js graphs with clickable nodes, descriptions, and prereq chains
  • 5+ React/FastAPI flows for major search, profiles, requirement views, and grade-difficulty data
  • Python scraping pipeline + MongoDB schemas across 1,000+ UCLA course records
ReactTypeScriptViteD3.jsFastAPIMongoDBFirebase Auth
Live Demo ↗Case Study

Project

Study-Strata

AI-powered academic planning platform with Redis-cached FastAPI backend

  • Full-stack platform generating personalized course schedules and degree roadmaps
  • Reduced API latency by 40% using Redis caching and asynchronous processing
  • Supports 100+ simulated student profiles with analytical scheduling features
ReactTypeScriptFastAPIPostgreSQLRedisDockerSupabase
Live Demo ↗Case StudyGitHub ↗

Project

MicroInvest

ML portfolio platform — 18% accuracy improvement across 4 forecasting models

  • TensorFlow and scikit-learn models with neural network and ensemble approaches
  • 15+ financial indicators compared across 4 forecasting model types
  • Streamlit dashboards for portfolio risk and performance visualization
PythonTensorFlowscikit-learnPandasStreamlit
Case StudyGitHub ↗