James Han
 November 2024 | |
| Contact | |
|---|---|
| jameshan.cs@gmail.com | |
| GitHub | @lxyhan |
| /in/jameshan27 | |
| @jameshan05 | |
| Page metadata | |
| First created | Nov 21, 2025 |
| Last edited | Jun 26, 2026 |
Hi, I’m James. I’m a 20-year-old student from Toronto. This wiki contains my research and engineering notes, plus writing across topics like programming languages, geography, and training for the National team in the Triathlon.
Currently
- Incoming at Wealthsimple’s data science and machine learning team.
- Retrieval and search infrastructure behind LLMs, Shopify, Sidekick + CX R&D.
- ML and networking research, Firefox privacy. Designed the methodology and the model: per-request bandwidth-cost estimation for blocked tracker requests, on-device ONNX inference, ~210M users. Coauthored the paper (under review); built the enhanced tracking protection metrics surface it feeds. Gecko security patches on the trust boundary.
- Built open-source software used by 30k+ students.
About me
- I enjoy ultras and triathlons! Training for the Canadian team for 2027 worlds, and racing a few 100km/80km trail races this year.
- I competed in debate internationally and coached students for four years.
- I’m a geography nerd. I have a collection of maps I find interesting (please ask me!!), and geek out on Geoguessr.
- I have a very, very cute cat.
Currently reading
- Friedrich Nietzsche, Thus Spoke Zarathustra
- Martin Heidegger, Being and Time
Built with Gleam, Rust, and Astro.
Index
- Summer 2026 Health Tracker
- Systems and ML at Firefox. Six months on the Firefox Privacy team: a tracker-cost estimation system, the multi-process data path in Gecko's C++ engine that feeds it, an offline training-and-distribution pipeline, and a new-tab privacy surface for 200M+ users. Core system scoped to landed in ten weeks; paper submitted to IMC.
- Search Infrastructure and Software Engineering at Shopify. Working notes from my engineering internship on Shopify's Sidekick and CX R&D team in Toronto, building the search infrastructure and software behind the help tooling around Sidekick. Mostly search and systems, with some applied LLM work. Written as I learn it.
- Machine Learning
- Open Source Contributions. Engineering on MarkUs and PythonTA, two open-source tools maintained at the University of Toronto and used across the CS department.
- Linear Algebra
- Triathlon
- Writing. Personal notes that I'm okay sharing publicly.