Hi, I'm JC, from the Philippines. 🇵🇭
I've always liked optimizing things. It started in high school, when I coded the color sensor for our team's robot at a robotics competition. These days I run five AI agents in terminal tabs, and most of my optimizing goes into knowing which one is waiting on me.
By day I write backend code at GCash. The rest of my time goes to AI that has to run somewhere specific: on a phone, on a laptop, or in healthcare software that can't send patient data to a model. I'm also working through a master's in AI at UT Austin.
At SapinSapin, which builds AI for Philippine languages, I train and evaluate Filipino speech-to-text and text-to-speech models on a free Colab GPU. I also lead development for CultureHub, an events platform for Australia's multicultural communities, with a team at Data Engineering Pilipinas.
A few things I believe:
- A small tool that works beats a big promise that doesn't.
- Constraints aren't a reason to wait. My local-model experiments run on the 8GB MacBook Air M1 I wrote my thesis on. It has never complained. It has every right to.
- AI should widen the door into this industry, not become another paywall.
- If I figured something out the hard way, I write it down for the next person, like your first open-source pull request.
I write at jcdiamante.com and go hiking. I also play the kalimba. It is the one thing I have not tried to optimize.