About the role
This is a co-founder role in substance: founder-level equity, founder-standard vesting, and shared risk. We are direct about that from the first conversation.
Kanan Labs builds the reasoning layer for cross-border trade. Every container that crosses a border is governed by a stack of rules no single person holds in their head — tariff lines, origin criteria, licensing conditions, each with a citation and an effective date, each changing without notice. We are building the system that holds them, reasons over them, and can always show its working.
The technical core is a compiler. It takes primary regulation, turns it into typed and versioned rules, signs them, and hands them to a product engine that produces evidence an exporter can put in front of an authority. Determinism matters more than cleverness here: an answer that cannot name its source, its version and its effective date is not an answer.
What this job entails. There is a constitution, an enforced type system, a working signing chain over one sealed slice and a green test suite. There is not yet a running product. Closing that gap is the job. For the first eighteen months you will be building, not managing — there is no team yet, and the work in front of you is code. Later you will hire and lead one.
You do not need to know anything about customs. Nobody arrives knowing this domain; the founder is a trade lawyer and supplies it. What you cannot learn on the job is the instinct that a rule which was true last year may not be true today, and that a system which cannot say "unclear" will eventually say something false with confidence.
What you’ll do
- Own the compiler end to end — typed primitives in, signed and versioned rule packs out.
- Make determinism real. Identical inputs produce byte-identical output, or there is a written reason why not.
- Build the signing and provenance chain so every output traces to the instrument it came from.
- Own privacy and data residency architecture — currently unowned, and the highest-severity item on the list.
- Choose and build the knowledge substrate, including how the system handles a rule that changes retroactively.
- Set the engineering bar, then hire against it.
What we look for
- You have built and maintained production systems where being wrong was expensive.
- Strong Python, and a preference for typed contracts over flexible dictionaries.
- You have worked with versioned data or schemas that had to stay correct as the underlying rules changed.
- You can explain why a fast non-deterministic answer can be worse than a slower deterministic one when the output has to be defended.
- You are comfortable being the only engineer for a while.
Nice to have
- Experience inside a standards-bound domain — healthcare terminology, financial reporting, tax, payments, identity — where the specification sat upstream of your opinion.
- Knowledge representation, ontologies, rules engines, or formal methods.
- Applied cryptography for evidence: signatures, canonicalisation, tamper-evident logs.
- Compiler or interpreter design. Bitemporal or versioned data modelling.
- Any exposure at all to customs, trade, tax or insurance regulation.