AI Engineer · Los Angeles

Shan
Yuxuan

I design AI systems

I love AI and I love building products people enjoy using. I design AI systems end to end, from LLM pipelines and agent harnesses to the cloud they run on and the interface people touch. When I'm not building, I'm out on a trail or on a stage with a mic.

scroll ↓
$400k+ARR generated
20TB+multimodal data pipelines built
3clouds I run: AWS, GCP, Azure
2.5 yrsto finish a 4-year NUS degree

01 What I do

AI systems

I design AI systems end to end: LLM pipelines that turn messy emails and documents into structured data, and agent harnesses that let assistants act safely through permission-scoped tools.

Product & UI/UX

I design and build products, from UX to production React and TypeScript. A good AI product should feel simple, even when the system behind it isn't.

Backend & infrastructure

I architect backends with FastAPI, MySQL and Redis, and run cloud infrastructure across AWS, GCP and Azure, from deployments to storage and background jobs.

Go-to-market

I like working close to customers: landing page analytics, marketing videos, onboarding, and following up with clients myself.

02 Experience

  1. Jan 2025 — Apr 2025

    Engineering Team Lead · Orbifold AI

    Silicon Valley

    • Built and deployed end-to-end LLM systems for retrieval, reasoning, and evaluation in production.
    • Trained multimodal embedding and voice separation models; built 20TB+ multimodal data ingestion pipelines.
    • Automated GPU inference with cost-aware scaling; built production APIs and customer-facing web apps.
    • Led client requirements and delivery, generating $400k+ ARR, and mentored interns.
  2. Jul 2024 — Nov 2024

    Research Assistant · A*STAR

    Singapore

    • Researched quantum machine learning, building quantum-classical hybrid models in Python and Qiskit.
  3. Dec 2023 — Jan 2024

    AI/Software Engineering Intern · Land Transport Authority

    Singapore

    • Tuned a YOLOv8 object detection application for traffic video and built batch video processing for it.
  4. May 2023 — Jul 2023

    Engineering Intern · NBE International

    Singapore

    • Debugged and validated PCB boards using Raspberry Pi, and built an automated monitoring system for fisheries.
  5. Oct 2018 — Jan 2019

    Research Assistant · A*STAR

    Singapore

    • Tested and benchmarked 12 signal compression algorithms for accuracy and efficiency using MATLAB.

03 Research

ICML 2026 · WorkshopsAgents · LLM evaluation

Towards Budget-Aware Agents: Do LLM Agents Know What They Will Spend?

Yuxiang Lin, Zihan Wang, Mengyang Liu, Yuxuan Shan, Longju Bai, Junyao Zhang, Xing Jin, Boshan Chen, Jinyan Su, Xingyao Wang, Jiaxin Pei, Manling Li

AI agents increasingly run under token, money and time budgets, but can they tell how much they're going to spend? This work defines budget awareness as an agent's ability to estimate, mid-task, how much budget it still needs and whether the task can still be finished. Frontier models turn out to be consistently over-optimistic and keep spending on tasks that are likely to fail. A budget-aware early stop saves 28–64% of tokens on failed runs, and the signal can be strengthened with training.

04 Recognition

1st place

Hackathon winner with Flair AI

Took first place with my team at a San Francisco AI User Group hackathon.

National award

Youth Corps Leaders Program

Received my certificate from then-President Halimah Yacob of Singapore at the Leaders Commencement Ceremony 2023, after leading Team Genesis.

05 Off the clock

Two things recharge me: mountains and music.

On the trail

I love to hike.

Some of my best thinking happens on a trail, far from a screen. A long climb has a lot in common with building a product: pick a summit, take it one switchback at a time, and enjoy the view when you get there.

trailheadsummitback down

Trail log

  • Denali, AlaskaUSA
  • More summits coming soon…

On stage

I love to sing.

From jazz sets with the NUS Engineering scholars to Mandarin classics, I'll take a mic and a live band any chance I get. Performing is a lot like demoing a product: rehearse hard, then trust it and enjoy the moment.

♪ Now playing: whatever the band counts in

06 Education & leadership

Jan 2025 — Dec 2025

Stanford University

Coursework in Management Science and Engineering (MS&E)

Aug 2022 — May 2026

National University of Singapore

Computer Engineering, First Class Honours

  • NUS Engineering Scholarship: finished 4 years of modules in 2.5 years
  • Teaching Assistant, University Ambassador
  • Chinese University Study Award

Jan 2024 — May 2024

Peking University

Exchange semester

Nov 2022 — Dec 2022

Korea University

Winter school

Global UGRAD Scholarship

Fully sponsored by the U.S. Department of State; one of only two students selected from Singapore each year.

Nov 2021 — Jan 2023

Team Leader, Team Genesis

Youthcorp Leaders Program. Recognized with a national award presented by President Halimah Yacob of Singapore. See photo ↑

Jul 2023 — Jul 2024

President, NUS Engineering Scholarship Club

07 Toolkit

AI/ML
AI System Design · LLM Pipelines · AI Agents · PyTorch · HuggingFace · Fine-Tuning · Quantization
Backend & Cloud
FastAPI · SQLAlchemy · MySQL · Redis · Docker · CI/CD · AWS · GCP · Azure · Terraform · Nginx
Product
Product UI/UX Design · React · TypeScript · HTML/CSS
Languages
Python · TypeScript · C/C++ · Java · R · SQL · MATLAB
Automation
Playwright · Selenium · Portal Reverse-Engineering

Let's build something good.

I'm always happy to talk about AI systems, product design, a trail worth hiking, or a song worth singing.

shanyuxuan2001@gmail.com
LinkedIn ↗