builder · hacker · shipper

I build things at the intersection of AI, engineering, and relentless curiosity.

Independent research, local-first AI benchmarks, and tools that solve real problems. Currently deep in the weeds with Gemma 4 on consumer hardware.

View Gemma 4 Benchmarks

Latest

Research

Gemma 4 Benchmark Suite

403 tests across 4 models (E2B → 31B) running on a 2017 iMac i7-7700K. Performance, reasoning, coding, tool-calling, creative writing, and agentic tasks — all benchmarked locally via Ollama. Real hardware, real numbers.

403 tests|4 models|10 categories

What I'm thinking about

Local AI

What can consumer hardware actually do? Not what the benchmarks say — what does it feel like to use a 31B model on a 5-year-old machine?

Agentic Workflows

Building with AI as a collaborator, not a tool. Pushing the boundaries of what's possible when you let the models drive.

Shipping > Talking

Less discourse, more output. The best way to understand something is to build it, break it, and build it again.