Thomas Lee
I like building trading systems and AI models
I'm a CS Specialist at the University of Toronto, building AlphaStone, an AI-assisted quantitative research project. I've worked across aerospace, enterprise IT, and academic research, won three hackathons, and currently research sparse attention mechanisms at UTMIST.
My long-term goal is to establish a research lab at the intersection of neuroscience and AI. I write about why current AI paradigms won't lead to AGI, and what neuroscience can contribute on X.
Experience
- Sep 2026 — Now
Jr. Product Manager, AI
WOLF Advanced Technology
Software Engineer Co-op
May–Dec 2025; Apr–Sep 2026
I built Blazor, C#, and ASP.NET Core tooling processing 18GB/day; delivered MES, NCR, CIS, QIR, and gap-pad validation systems. Improved manufacturing and quality workflows across WOLF's full product portfolio, including the WOLF-3180 FGX2 module used in NASA's X-59 XVS; increased throughput 1.3× and reduced quality errors 41%. Refactored a 26-year-old VB6 RMA system, resolved a production-critical website crash within two weeks, and supported C++ firmware testing and Python development for an AI self-driving JetTank.
Best Employee of the MonthReturn offer · U.S. relocation + immigration support - Aug 2026 — Now
Engineering Project Manager
UofT Machine Intelligence Student Team
ML Researcher
Jan 2026 — Aug 2026
Promoted to manage internal ML and industry projects, roadmaps, and multiple external stakeholders. Conducted R&D on Bigger Bird, a PyTorch/Triton sparse-attention router reducing complexity from O(n²) to O(n); fine-tuned BART and R1 on H100 GPUs, reducing the complexity slope from 1.65 to 0.70 with 0.88 F1.
- Jun 2021 — Now
Founder
AlphaStone · Quantitative trading platform
I lead quantitative trading R&D while managing $70K+ AUM across six investors with $2K–$4K monthly profit. I built the full research and trading stack in Python, C, Node.js, tRPC, PostgreSQL, and Nuxt/Vue for execution, accounting, dashboards, and backend services. The platform processes millions of multi-timeframe events through cached, event-driven backtesting and risk infrastructure with realistic execution and cost modeling.
- Sep 2024 — Apr 2025
Research Software Engineer
UofT — Division of VP, Research & Innovation
I built Python and C data pipelines on Azure SQL Server to power UTQAP statistics, big-data analysis, and Tableau dashboards supporting DiscoverResearch's 2,600+ public faculty profiles, including Nobel laureate Geoffrey Hinton. I also developed Next.js, Flask, and C++ tools for the Research Security initiative.
Return offer - May — Dec 2024
Full-stack Developer Co-op
Softchoice Corp. — IT solutions for MS, Google, AWS, Cisco
I built a CRUD web app for test leads to schedule and manage QA tests, linking DevOps API results, reports, and QE permissions. I owned 50 REST endpoints across SQL, MSAL, JWS, and Azure DevOps, set up 2 Azure VMs and YAML pipelines, and wrote Java Selenium test automation suites. I pitched migrating from React/SPFx to Nuxt/Vue.js.
Return offer - 2022 — Present
University of Toronto, St. George
Honours BSc in Computer Science — Focus: Artificial Intelligence
Projects
Aircraft Studio
Mobile web app that designs 3D aircraft in seconds and simulates against enemy jets in AR.
MindAssist
EEG-controlled robotic arm for hands-free eating. Decoded brain signals via Bluetooth to drive a 5-DOF arm with computer vision.
Milestones of Humanity
An interactive data visualization of humanity's greatest milestones — from the Stone Age to the printing press to the moon landing.
TrashCam
Real-time object detection for automated trash sorting. Next.js, COCO, Google Cloud Vision, Gemini.
MagicQuill
AI note-taking with speech, image, audio-to-text, flowcharts, and LaTeX rendering.
JARVIS
Speech-to-anything AR experience with agentic capabilities — ordering food, processing video, all in a browser.
Forecast AI
Led a team of 7 under Stanford/UofT researchers to build a forecasting research project at the Department of CS ML Group.
Media
FreeCodeCamp released a documentary covering TreeHacks 2026
Here's MindAssist demo I did for FreeCodeCamp's documentary covering TreeHacks 2026 at Stanford.
IMI Big Data and AI Competition
Brief reflection after participating in the IMI Big Data and AI Competition leading a team.
If you're working on something interesting — neocortex research, quant ideas, or just something crazy — I'd love to hear about it.
yehyun.lee@mail.utoronto.ca