"I'm sorry, Dave. I'm afraid I can't do that."

HAL 9000, 2001: A Space Odyssey

Science fiction got one thing badly wrong. The interesting problem was never the machine that refuses. It is the machine that agrees, confidently, at scale, with your money. This is the company building the part where a qualified human gets to look at the answer before anyone acts on it.

The story

Financial services has a supply problem that nobody has solved by hiring. There are not enough qualified professionals, the good ones are expensive and slow to reach, and most people end up with either a rushed answer or no answer at all. Software was meant to fix this and largely did not, because the existing tools make the customer do the work and then cannot answer the one question they actually came with.

This business took a different run at it. An AI agent does the work end to end, and licensed professionals review every high impact decision before it goes anywhere. The review loop is the product, not a disclaimer at the bottom of it. Every correction a professional makes becomes signal, the agent gets sharper, and the human time moves up the value chain instead of being burned on the routine.

Most AI in this sector is sold business to business, as software that makes professionals faster and margins better. This one is pointed straight at the end customer. That changes what you build, because an internal tool can be clunky and a consumer product in a regulated market cannot.

Why now

Seed stage. Around ten people, roughly half of them engineers. The product is live, customers are paying, and the architecture is still genuinely unfinished.

That last part is the whole pitch. The services are not all named yet. The event model is not locked. The patterns another engineer will be reading in three years have not been written. Join a year from now and you inherit all of it. Join now and you are one of the four or five people who decided it.

The founding team has done this before, including taking a company from seed to a public listing, so the ambition is not theoretical. Backing comes from operators inside two names in AI and fintech you would recognise immediately. Equity is real, and at this headcount it is early enough to matter.

The work

You own features properly. Data model, services, API, interface, and then running the thing in production once it is live. No throwing it over a wall, because there is no wall and nobody on the other side of it.

The product layer is TypeScript end to end. Node on the backend, React or Next.js on the front. This is where most of your week goes, and it is why production TypeScript is the one thing nobody gets waved through on.

Underneath it sits the agent, and that lives in Python. You are not being hired to build models or write evals, there is a founding AI engineer doing exactly that. You are being hired to build the system around it. Calling agent services, handling the responses, designing the failure paths, deciding what happens when the agent is slow or uncertain or wrong, and building the interfaces that let a professional review, override and sign off without friction. Comfortable reading and shipping Python is the baseline. You will spend real time on that boundary.

Then the event layer. Kafka or equivalent, because the work is asynchronous by nature. Documents arrive, the agent processes, a human reviews, state advances, the customer gets told. That flow has to be reliable, replayable and idempotent, and in a regulated market the event log doubles as the audit trail. Getting this right early is the difference between a system you can explain to a regulator and one you cannot.

The regulatory constraint is not paperwork, it is architecture. Auditability, access controls and careful handling of sensitive data get designed in from the first commit, not retrofitted the week before a review. If you have built anything in payments, billing, ledger or risk, you already know why that matters and how expensive it is to bolt on afterwards.

The team is AI native. They use AI coding tools every day, and whoever joins helps set how that is done properly rather than just enthusiastically.

Stack

TypeScript, Node.js, Python, React, Next.js, Kafka.

Essential. Please read this part before applying.

This is a narrow brief and we are screening tightly against it.

• Production TypeScript. Node.js on the backend, React or Next.js on the front. Shipped and running, not learned on the side.

• Python services in production. Built and operated, not just read.

• Kafka or equivalent event driven pipelines. Real production event flows.

• Distributed systems and microservices. Designed, not only worked inside someone else's.

• Fullstack end to end. Features taken from data model through to UI and then run in production.

• Five or more years as a full time engineer since finishing study.

• Early stage experience. At least one startup, or a startup like team inside a larger business.

• Degree in computer science or a related STEM field.

Strongly preferred

• Payments, billing, ledger or risk systems built at a fintech.
• Security and compliance work: sensitive data, auditability, access controls.

Not a fit

Contractors. Big company only backgrounds with no early stage exposure. Python specialists without production TypeScript, and TypeScript specialists who have never shipped a backend service.

Practical details

Fully remote within the US, for candidates based in New York, Chicago or the Pacific time zone. Quarterly offsites, no relocation. US citizens and Green Card holders only, no visa sponsorship.

Senior Software Engineer | TypeScript, Python, Kafka

AI and financial services. Remote, US. Up to $225K base plus equity.