Maxwill Lin
Maxwill Lin

Current Work

At Vmax, I designed and built the core system end-to-end for training software-engineering agents: RL infrastructure that orchestrates hundreds of GPUs, a self-improving task generator, and a rollout framework that runs thousands of concurrent sandboxes; I am actively contributing upstream across slime, Miles, and Harbor.

Earlier: Meta MRS (Meta Recommendation System), autopilot systems, cryptography/blockchain protocols, and quantitative research.

Professional Experience

  • Vmax – RL infrastructure and self-improving task generation Designed and built the core system end-to-end: an internal slime fork, a self-improving agentic RL environment generator, and a rollout framework for thousands of concurrent sandboxes; actively contributing upstream across slime, Miles, and Harbor.
  • Meta – MRS (Meta Recommendation System) Applied ML Researcher working on LLMs for ads recommendation.
  • Previous roles Autopilot systems · Cryptography & Blockchain protocols · Quantitative research.

Education

  • M.S. Computer Science, Georgia Institute of Technology — Algorithms & Generative ML
  • B.S. Computer Science & Applied Mathematics, National Yang Ming Chiao Tung University — Theoretical CS & Mathematics (Top‑2 CS program in Taiwan)

Some Facts

  • Top 0.56% of a 96,000+ registrant field in Google Code Jam 2020
  • ~USD 9k prize at ETHGlobal Paris (1,400 participants, 321 projects)
  • I enjoyed challenging myself with theoretical CS/ math courses (e.g. Quantum Computation and Quantum Information, Quantum Machine Learning, Advanced Algorithms and Uncertainty, Intractable Problems, Advanced Linear Algebra, Algebra)

Beyond Work

I grew up in Taipei, Taiwan (Mandarin name: 林彥彤 / Yan‑Tong Lin).

Hobbies

Web3 Addresses

0xb1B6356EA9E2f3Bf9867d6Ac1c1Bfd2cB1553Abb  
ytlin.eth  
maxwill.sol