a software developer from Taiwan, currently with Chunghwa Telecom. I’m not entirely sure if I should label myself as an “AI engineer” since it’s such a fresh field. I see myself more as someone doing software engineering and integration in the AI space. While my role is somewhat akin to that of a software architect, I don’t feel I’ve earned that title yet. The more I learn, the more I see how much I still have to grow, and there's never enough time.
In the past, I was really into hands-on coding, but by 2025, I’ve slowly shifted to being a “conductor,” focusing on guiding AI to write code. For instance, with bgutil-ytdlp-pot-provider-rs and subx-cli, I hardly did any coding myself during their development; instead, I directed AI to handle it. I've been on a journey to level up my vibe coding to production quality, picking up some techniques along the way which now known as Spec-Driven Development. It wasn’t until this term came up that I realized everyone was pretty much doing the same thing.
Besides the usual software engineering stuff, I'm mainly diving into AI field particularly in advanced prompt engineering, or recently people have started calling it context engineering. I've also written a bunch of blog posts on this subject. The standout ones are "Say Goodbye to AI Hallucinations: A Simple Method to Improving the Accuracy of Your RAG System" and "🤖 AI can cosplay too? A beginner's guide to LLM character role-playing." You can check out all my AI-related writings on my blog.
I spent four years working as a professional C# .NET developer, but just this year, my output in Python and Rust has already surpassed my C# work. It's not that I've stopped loving C#, but these two languages have really thrived thanks to the AI boom. Even so, my primary tech stack is firmly rooted in C#. Feel free to explore my C# repositories and my (biggest) side project Recorder-moe.
I'm pretty well-versed in containerization technology and have a knack for creating advanced containers. I can create images that are well-designed, prioritize security, minimal, highly reusable layer caches, and following industry best practices. I have spent a lot of time and effort on this technology, and I believe that what I produce is top-notch. You can find those cool stuff I've worked on in my Containerfile-template repository.
I have some fundamental understanding of kubernetes as well. In my home lab, I'm operating a cluster using bare-metal Talos Linux for my personal services. Building it was a real challenge, and I learned a ton throughout the process. I've completed an OpenShift course and have even done deployments in a corporate setting. While I didn't end up getting certified, I'm confident in my ability to get the job done.
I heavily rely on GitHub Workflows for automating tasks in my open source projects. I also utilize Azure Pipelines and GitLab Pipelines at work and have experience in setting up agents/runners and customizing workloads. In essence, I am able to setup CI/CD across all three platforms. Check this if you want some true example.
Oh, and I am fully familiar with DevOps and GitOps, which seems to go without saying.
I gotta talk about cloud computing services. I have delve into both Cloudflare Workers and Azure Functions. They are truly a blessing for those on a tight budget! Thanks to them, I managed to slash the operational costs of Sound Buttons to practically nothing on the backend servers.
Furthermore, Cloudflare workers are great for generating open graph meta tags in front of the SPA and also well-suited for web applications that experience high traffic. I have written a few blog posts on this topic in Chinese.
More info
Feel free to find me here, and I mainly lives in Mastodon.