I’m Matt Ngaw, an engineer interested in computer architecture and systems. I currently do research at Arm on scaling machine learning workloads in the datacenter. I recently earned my B.S. and M.S. in ECE at Carnegie Mellon!
You can find me on Github, LinkedIn, and X
What I’m up to now
I’m currently digging into the implementation of llama.cpp, an LLM inference engine geared towards consumer hardware. I am starting with GGML, its underlying tensor library. During the day, I work on ML in the cloud, but long-term I care about ML on consumer hardware!
What this site is for
See Learning After College for my motivations for this site.
What’s in my homelab
I’ve currently got the following hardware:
- An RTX 3090 Ti, my main driver for experiments and local inference.
- A Raspberry Pi 5 + an RTX 3060, connected over Oculink.
- A gaming laptop running protein folding 24/7.
- An 8-core Sandy Bridge-EP Xeon with 2x RX 480s.
- A Jetson Nano (4GB).
- Various other Raspberry Pi’s.