LLM Lab
I built LLM Lab to compare prompts, run repeatable evaluations and track model costs. The next iteration brings datasets, scorers and experiment traces into a shared evaluation workflow.
Architecture preview ↗Intelligent Iterations
The products and engineering tools we’re building at ii.
I co-founded Intelligent Iterations and lead its technical work, from application development and model evaluation to the infrastructure that runs our agents.
Our current platform
A workspace for AI generation, media editing and coding agents, connected to infrastructure the customer controls.
Explore ii-osI built LLM Lab to compare prompts, run repeatable evaluations and track model costs. The next iteration brings datasets, scorers and experiment traces into a shared evaluation workflow.
Architecture preview ↗Ingredient scanning and research data. I built the data pipeline with ML Kit and batched vLLM inference, delivered versioned datasets, and trained an OCR model that runs on the device.
My role and experience ↓The compute layer behind ii-os. It assigns jobs to available machines, runs them in disposable containers or VMs, and verifies cleanup before releasing capacity.
How it fits into ii-os ↗Tools for working with coding agents from Slack. The harness connects Claude Code and Codex sessions to a shared interface for starting work and following its progress.
Public repository ↗More from ii on GitHub ↗