We're hiring
Build the verification layer for consequential AI
Kenshiki verifies AI outputs against authoritative evidence before anyone can act on them. Models are good at generating plausible answers. We make sure those answers are provably correct — or they don't ship.
Principal AI Taxonomist / Knowledge Engineer
Define the evidence rules that control what reaches the model — translate regulatory frameworks into the structured Kura data that makes our governance gates work.
Principal Data Reliability Engineer (AI Process Control)
Own the quality math inside the bounded-synthesis pipeline — define the tolerances, build the telemetry, and make sure every governance gate has a measurable threshold.
Principal Systems Engineer (TypeScript & PostgreSQL)
Build the runtime — the database schema, hybrid search, and governance middleware that sits between evidence and the model.
Lead Forward-Deployed Infrastructure Engineer (Air-Gapped Systems)
Take our stack across the air gap — package, deploy, and maintain sovereign Kenshiki appliances on bare-metal hardware inside secure enclaves.
Platform & Ingestion Engineer (Data Ops & Infrastructure)
Own the seams — the evidence creation pipeline that turns raw documents into verified evidence, and the build system that packages everything for deployment.
How We Work
Operating Constraints (Not Brand Language)
Our core values — Truth Over Fluency, Authority Is Earned, Refuse When You Should, and Build the System, Not the Fix — are the structural axioms of our architecture and our company. If you do not fundamentally agree with them, you will hate working here.
Our Focus (The Hedgehog)
We operate on Jim Collins' Hedgehog Concept. We ignore the broader AI hype cycle to focus entirely on building deterministic enforcement architecture for generative AI. We do not build generic wrappers; we build the cage.
The No A**hole Rule
Kenshiki is a small team where everyone ships across boundaries. We balance our technical ambition with strict adherence to Bob Sutton's No A**hole Rule. We do not hire brilliant jerks. You can be the most talented systems engineer on the planet, but if you demean your peers, hoard knowledge, or lead with ego, you have no place here.
The Monday Morning Reality
The pace here is measured but relentless. We don't do artificial sprints; we do deep, concentrated engineering. You will be writing high-performance TypeScript, optimizing complex PostgreSQL/pgvector schemas, and squeezing every drop of performance out of local inference engines for air-gapped deployments. Quality is structural. We expect you to leave the codebase better than you found it.
How We Interview
We look for people who are introspective, humble, and wicked smart. Expect us to ask about a time you failed catastrophically at work due to a decision you made. We want to hear how you dissect your own wreckage, own the blast radius, and re-engineer the system so it never happens again.
Note: If you have been fired in the past for taking a risk that went south, a system failure, or a culture mismatch — tell us. We view operational scar tissue as a badge of honor, not a red flag.