Co-founder, MLNavigator

James KC Auchterlonie

Building adapterOS · Verifiable AI Infrastructure

I build reproducible ML runtimes — same input, same output, with a record of what ran. Designed for industries where you need to show your work.

Companies & Projects

Deterministic AI tooling for regulated industries.

AI research group building deterministic inference tooling. Co-founded with Donella Cohen. NSF I-Corps cohort, Kansas ACCEL grant recipient.

Deterministic inference runtime for Apple Silicon. Same input, same output — with a record of what ran. MLX and Rust.

What I spend my time on and why it matters.

Same input, same output. Seeded randomness, pinned dependencies, serialized pipelines — so you can replay any run.

Execution receipts that record what was computed, when, and with what configuration. Built for audit requirements.

High-performance Rust and MLX on Apple Silicon. Low-level optimization for inference runtimes and LoRA adapter orchestration.

Targeting healthcare, finance, and legal — industries where showing your work isn't optional. On-device inference for privacy.

On building verifiable AI, navigating early markets, and systems that survive scrutiny.

Fast, local-only utilities. No tracking, no accounts.