我之前主要做三维视觉算法，包括 SLAM、点云、相机位姿优化、三维重建。
一开始，我关注的是将算法理论进行工程化落地与产品化呈现。在 AI 这波浪潮席卷下，我逐渐关心一个更重要的问题：
我的想法，以及生活中的一些很具体的小问题，怎么样才能变成别人一打开就能使用的产品？
所以我开始尝试一些小工具和产品。它们可能非常小，但解决的问题非常具体。
于是我也就有了新的轨迹：
充分利用 AI 的实现能力，将想法落实到一个个具体的事件上。
凡所思，皆可造。
English
I used to work mainly on 3D vision algorithms, including SLAM, point clouds, camera pose optimization, and 3D reconstruction.
At first, I focused on turning algorithmic theory into engineering work and product-facing results. As AI became more powerful, I started to care about a more important question:
How can my ideas, and some very specific problems in daily life, become products that people can open and use right away?
So I began building small tools and products. They may be very small, but the problems they solve are very specific.
This became a new direction for me:
using AI’s implementation ability to turn ideas into concrete things, one by one.
These are the projects that best explain what I’m building now.
01. AIGSplat
A web-based 3D Gaussian Splatting and point-cloud viewer.
This is the project closest to where I came from: 3D reconstruction, spatial data, and neural rendering.
I want reconstruction results to be easier to view, share, and use — not just stay inside training scripts or research demos.
02. AhaChat
A local-first AI chat workspace for topic-based threads, reusable roles, and OpenAI-compatible providers.
I built it because normal AI chat often becomes messy very quickly.
Different topics, roles, prompts, and model providers need a more structured place to live.
03. AhaMark
A browser-based Markdown workspace for writing, preview, outline, export, and local drafts.
This is a small writing tool, but it reflects something I care about:
turning scattered thoughts into structured output.
I use it as part of my own workflow for writing, documentation, product notes, and idea organization.
04. Pauza
A quiet desktop break reminder.
I didn’t want another noisy productivity dashboard.
Pauza is a small attempt to make a tool that reminds you to pause without making “wellness” feel like another task.
I also keep a small product lab under AhaKnow.
Some projects are iOS apps.
Some are web tools.
Some are AI utilities.
Some are just experiments that helped me test an idea quickly.
They are not all meant to become big products.
For me, they are a way to practice the full path:
notice a problem → define a small version → build it → ship it → learn from it.
- SayCam — a quiet self-recording camera for daily video journaling.
- 天時 — a weather and classical poetry app.
- RealAlpha — turns fake-transparent AI-generated images into real alpha-channel PNGs.
- AddressGO — AI-powered address translation app for cross-border forms, shipping, visas, and banking.
- Icon — an oriental-style icon workshop for app icons, color palettes, seals, and textures.
- AhaPPT — a folder-driven presentation site for image-based decks.
- AhaTTS — an OpenAI-compatible text-to-speech service.
- StillHere — a quiet life-confirmation app.
I’m currently focusing on three things:
Better ways to represent, view, and use reconstructed spaces.
Using AI not only to write code faster, but to move from unclear ideas to working software.
Products with real URLs, simple operations, and clear use cases.
