James KC Auchterlonie builds systems for regulated and offline AI deployment. He is the creator of AdapterOS, a runtime for deterministic inference and cryptographically verifiable execution receipts. His work focuses on token-level audit trails and canonical serialization. Based in Wichita, he bridges the gap between engineering reality and procurement compliance for government-facing technical projects.

Current status: Recent patent filing for deterministic inference receipts.

Technical Themes

  • Determinism as a primary product requirement.
  • Design of cryptographically verifiable execution receipts.
  • Focus on token-level audit trails.
  • Canonical serialization as a first-class constraint.
  • Modern hashing and derivation: BLAKE3, HKDF-SHA256.
  • Fixed-point / Q15 constraints to eliminate numeric drift.
  • Accountable adapter routing and MoE gating decisions.
  • Priority for offline and regulated deployments.

Implementation Posture

  • I build runtimes. I do not just train models.
  • Inference must be replayable and evidence-backed.
  • Version-locked artifacts and reproducible builds are standard.
  • Strict environment separation: dev, prod, and archive.
  • Full-stack engineering: Rust, Axum, and infrastructure-as-code.

Operational Capability

  • Homelab-to-production deployment workflows.
  • Automated infrastructure management via Terraform.
  • Local proxy layers and Cloudflare tunnel ingress.
  • Audit-ready systems mapping to procurement requirements.

Strategic Motion

  • Patent application filed for deterministic inference receipts.
  • Direct focus on government-facing applied research.
  • Narrative centered on regulated trust over consumer hype.
  • Technical translation between engineering and compliance.

Style

  • Explicit structure prioritized over general aesthetics.
  • Traceable reasoning and deterministic outputs.
  • Systems built for hostile verification.
  • Anchor-audit-rectify operational loops.

Timeline

Founder

MLNavigator Inc.

Founder

AuchShop LLC

Textron Aviation

2019-2022

Bombardier

2023

CTI

2024-2025