English | 简体中文
Gee-eS-yoU-aI-fifty-fifty-one, or Gee-eS-yoU-aI-five-zero-five-one.
If you wish, you can even read it as Golf Sierra Uniform India five zero five one. That's also accepted.
By day, I'm an engineer at an airline, with an aerospace background and a lasting fascination with aerodynamics. Look into where physics, data, and judgment meet.
Outside of work, I'm an independent builder interested in open source, web applications, geospatial data, telemetry, outdoor activities, and AI-assisted development.
I like turning things that are scattered across maps, devices, apps, and data files into tools that are easier to explore, measure, and understand.
My interests tend to sit somewhere around:
Maps × Data × Hardware × Human Activity
Telemetry for every way. Sliced.
An open-source, web-based outdoor activity data analyzer built around sector-based analysis and telemetry.
Keywords:
- 🗺️ GPX / KML / KMZ / TCX / FIT parsing
- 📏 Distance and 3D distance
- ⏱️ Time, pace and sector analysis
- ⛰️ Elevation gain/loss and slope
- ❤️ Heart-rate zones and telemetry
- 🚴 Cycling / 🏃 Running / 🥾 Hiking
- 🧭 Multiple map sources
- 🌍 Metric / Imperial units
- 🌐 Multi-language interface
- 💻 Web-based and local-first where possible. Your data, only yours.
Instead of treating a recorded activity as just a line on a map, WaySlice treats it as a collection of measurable sectors with visualized telemetry.
I have a strong interest in:
Maps × Data × Hardware × Human activity
Recurring areas of interest:
- 🗺️ Geospatial and route data
- 📡 Telemetry and sensor data
- ⛰️ Hiking and outdoor navigation
- 🚴 Cycling and endurance activities
- 📊 Data visualization and analysis
- 🌐 Small, focused web applications
- 🧩 Open-source software
- 🤖 AI-assisted software development
I'm particularly interested in tools that answer questions like:
Where did I actually spend my time?
What happened on this section of the track?
How did elevation gain/loss affect my effort?
What can the raw telemetry tell me that the original app doesn't?
Not like：
Where is the track?
I tend to choose tools pragmatically rather than following a fixed stack — I'd rather evaluate fit, long-term cost, reliability, and ecosystem than default to whatever is popular.
JavaScript · HTML · CSS · Leaflet · SVG
GPX · KML · KMZ · TCX · FIT · GeoJSON
Git · GitHub · Python · Node.js
LLMs · Coding Agents · Prompt Design
One of my current interests is how AI is changing the way software is built.
I experiment with different:
- Coding agents
- Different LLMs for different tasks
- API providers
- Intelligence vs Latency vs cost
- Development environments with LLMs
- Long-context workflows
- File-and-content-based workflows
- AI-assisted translation and documentation
- Prompt design
I also like benchmarking models on the tasks I actually care about: translation quality on domain-specific text, image understanding, reasoning, and whether a model gets obscure cultural references. I'd rather do that than trust leaderboard scores.
I'm particularly interested in the trade-offs between:
Intelligence × Latency × Cost × Context × Reliability
I don't believe there is a single "best model".
The more interesting question is:
Which model is on the efficient frontier for a specified task?
Telemetry for every way. Sliced.
所经之路，皆成遥测；所成遥测，皆可切析。
Open-source activity analyzer focused on sector-based telemetry.
Maps · Tracks · Telemetry · Data Visualization
An earlier exploration into browser-based, local GPX analysis.
TrailScope helped shape many of the ideas that eventually became WaySlice.
An Incremental/idle game on making antimatter, reaching infinity and more.
Top translator for Simplified Chinese in mobile (Android and iOS) version.
→ Android
→ iOS
→ Chinese Version based of my translation
An Incremental/idle game about a fish-eating shark and more.
Top contributor for Simplified Chinese text.
When I'm not building software, I spend a lot of time walking, hiking, trail running, and cycling.
I particularly enjoy long-distance trails and routes where the journey itself produces interesting data.
A few other corners of my brain:
- ✨ Stargazing not just going outside, but going for astronomy phenomena. Such as solar/lunar/stellar transit of International Space Station or Tiangong Space Station, lunar occultation, solar/lunar eclipse, annular solar eclipse (21 June 2020 in Amoy), venus transit (6 June 2012)
- 🏎️ F1 & race strategy aerodynamics, tire windows, undercuts, and the engineering of going faster
- 🌦️ Weather models comparing numerical forecasts before big outdoor days; basically a one-person pre-event weather briefing
- 🗣️ Languages native in Mandarin and Cantonese, working in Mandarin and English; endlessly curious about translation logic and how words get transliterated
- 🎧 K-pop every long climb and descent, deserves a soundtrack
That is also part of the reason WaySlice exists:
If I've already recorded the journey, why shouldn't I be able to understand it?
- 🐰 Jellycat bunnies (the bashful kind)
- 🕐 Once spent way too long figuring out which world-clock city best for UTC simulation with these conditions: UTC+0 all year round, no DST ever, as close to the Prime Meridian as possible
Tools should expose the data, not hide it.
A map is more than a background layer.
Telemetry becomes useful when you can slice it into meaningful sections.
The best tool is often the smallest tool that solves the actual problem.
AI changes software development, but it doesn't remove the need to think about software development.
- GitHub GSUI5051
- Bilibili 少女们5051
- Zhihu GSUI5051
- Weibo GSUI5051
- Steam Fanytastic
- YouTube Fanytastic
- Twitter / X GSUI5051
Or send an E-mail to gsui5051 (at) qq (dot) com