Yin-Chi Chan

Top Spin (revisited)

Fri, 15 May 2026 00:00:00 +0000

In a (much) earlier post of mine, I discussed a puzzle called Top Spin (once made by Binary Arts), in which 20 beads were placed on an oval track with a turntable that could flip the four beads that occupy it. I showed that it was possible to reverse the order of the beads, from [1, 2, 3, ..., 19, 20] to [20, 19, ..., 2, 1], even if the beads were placed on a straight track instead of an oval one.

More Thoughts on CodeCrafters

Sun, 10 May 2026 00:00:00 +0000

In my previous post, I shared some of my thoughts on the CodeCrafters website and the coding challenges it offers. In particular, I had completed the HTTP server challenge in Rust and Claude Code challenge beta (notably, the number of available stages, and the feature set of the “completed” code, are quite limited) in Python and was working on the Interpreter challenge in Rust. I have since completed the Interpreter challenge, and decided to share some more thoughts on the experience.

Thoughts on CodeCrafters

Sat, 04 Apr 2026 00:00:00 +0000

CodeCrafters is a site where one can complete a number of programming challenges. Challenges include:

Build your own HTTP server
Build your own Interpreter
Build your own Shell
Build your own Claude Code (still in beta)

Each challenge is broken into a number of steps, with an extensive test suite triggered by the command codecrafters submit which:

Pushes your submission to a Git repo on the CodeCrafters server
Runs the tests for the given step
Reports results back to the user in the terminal

Supposedly, this is made possible by Git’s post-receive hook, which runs the test suite and reports the results back to the user via Git’s side-channel communication capabilities (as well as updating the user status in Codecrafter’s database if the step was cleared sucessfully).

A flip puzzle

Wed, 09 Jul 2025 00:00:00 +0000

Top Spin was a puzzle in which 20 beads were placed on an oval track with a turntable, such that any four consecutive beads could be flipped. Naturally, I wondered whether it was possible to reverse the order of the beads, from [0, 1, ..., 19] to [19, 18, ..., 0].

As it turns out, this is possible even if the beads are placed on a straight track instead of an oval one.

Automating my publications page

Sat, 19 Apr 2025 00:00:00 +0000

This post refers to a previous version of my personal website, which was built using MkDocs. The current version of my website is built using Hugo, and the content of this post may not be relevant to the current version.

A referenced Python package below, mkdocs-bibtex, is no longer maintained. In my new Hugo-based website, this package is not necessary — the Taho theme for Hugo providing a built-in solution for displaying publications from a YAML file, which is generated using pybtex directly (mkdocs-bibtex used pybtex as a dependency).

About Me

Mon, 01 Jan 0001 00:00:00 +0000

Background

I was born in Hong Kong and moved to Canada for several years as a child, before returning to Canada to study a BMath at the University of Waterloo. I then completed a Masters and PhD at the City University of Hong Kong. In Dec 2021, I moved to the UK to work as a research associate at the Institute for Manufacturing, University of Cambridge. I’m currently based in the Toronto area, seeking industry roles in data science, simulation, or backend/data-focused software engineering.

Projects

Mon, 01 Jan 0001 00:00:00 +0000

Command-line LLM chatbot demo

https://github.com/yinchi/openai_demo

This simple chatbot uses the OpenAI Chat Completions API to generate responses to user input. Responses are interpreted as Markdown and rendered in the terminal using the rich Python library. Available tools include file read/write, Python code execution, and web search via the Brave Search API.

The chatbot was tested locally using a vLLM instance running the google/gemma-4-26B-A4B-it model. A gemma.sh script is included in the repo to simplify the process of launching the vLLM instance with the correct parameters.

Publications

Mon, 01 Jan 0001 00:00:00 +0000