Software developer · Open source
I build practical tools for
local AI infrastructure.
Python, C++, and SQL. A focus on the systems that make local models useful.
One active backend at a time
01 / Selected work
Two open-source Python tools.
From routing requests to understanding runtime behavior.
01 Inference orchestration
One API for multiple local inference engines. Route by model, select for the task, and swap backends when GPU memory is tight.
Inside the projectOne active backend at a time
02 Runtime observability
An nvtop-style terminal monitor for local model runtimes. Discover engines, inspect GPU and serving metrics, and replay recorded snapshots.
Inside the project02 / About me
I'm Graham, a software developer focused on Python, C++, SQL, and open-source AI infrastructure.
My work sits around local model inference: making different engines work together, handling memory constraints, and making runtime behavior easier to understand. I created local-engine-router and llmtop to address those problems.
I also have experience with parameter-efficient fine-tuning using LoRA and QLoRA to automate enterprise workflows. I'm currently completing an Associate in Science in Computer Programming at Valencia College.
