About
I'm studying information technology at Macquarie University in Sydney, majoring in AI. Most of what I build sits close to the model: quantized inference on phones, interpretability experiments on small transformers, and evaluation harnesses that try to say something more useful than a single benchmark score.
I care about measurement. It's easy to ship something that looks like it works — a demo, a fine-tune, a leaderboard number. It's harder to show what actually changed and what it cost. That gap is what most of my projects are about.
Outside that, I build and fly FPV drones, run, lift, and follow Formula 1 more closely than is strictly reasonable.
Currently
- Building
- A quantization benchmark suite for on-device models — Swift over llama.cpp.
- Reading
- Circuits work on induction heads and in-context learning.
- Open to
- ML engineering internships and research assistant work.
Working with
- ML
- PyTorch, TransformerLens, llama.cpp, Hugging Face
- Build
- Python, TypeScript, Swift, Next.js, Supabase, Prisma
- Other
- Git, Docker, SQL, Linux
Get in touch
Email is best: you@example.com. I'm also on GitHub and LinkedIn.