AI Product Engineer

Verifast verifies identity, income, and employment for property management, mortgage, and KYC

clients across the US and Canada. A large amount of what we do involves judgment calls on messy

documents, edge-case applicant situations, and client-specific rules. Much of that work can now be

built rather than staffed.

We're hiring an AI Product Engineer to build those solutions. You will work primarily with Claude and

similar tools to turn client problems into working product: agents, automations, internal tools, and

customer-facing features.

This is a product role first and a building role second. You will own problems end to end: talking to

clients and internal teams to understand where the pain actually is, deciding what's worth building

and what isn't, shipping it, and measuring whether it worked. The building is what makes the product

work possible, not the other way around.

This is not a research role and it is not a traditional software engineering role. We care much more

about whether you can take a vague, messy business problem and ship something that actually solves

it than about whether you can pass a systems design interview. The best person for this job might be a

technical product manager who learned to build, a support or operations lead who automated their

own function, a founder whose startup didn't work out, or a self-taught engineer with unusually good

product instincts.

What you'll build:

These are real problems on our roadmap, not hypotheticals:

• Document intelligence. Pay stubs, bank statements, offer letters, tax transcripts, and benefits letters arrive in every possible format. Build extraction and validation that handles the long tail without a human touching it.
• Support automation. We run an AI support assistant and a Zendesk operation with tight SLAs.
• Build agents that resolve tickets end to end, enrich them with context from our data, and enrich correctly.
• Client-facing workflows. Leasing agents and property managers spend hours on repetitive review work. Build tools that compress that into minutes.
• Internal leverage. Reporting, reconciliation, QA, and data cleanup work that currently consumes analyst time. Automate it.
• Solutioning for clients. When a large client has a workflow problem that our standard product doesn't cover, scope it, prototype a solution, and get it in front of them quickly.

Responsibilities:

Product

• Own problems end to end, from discovery through launch and iteration. Often you will be the one defining the problem, not just solving it.
• Run discovery directly with clients, Client Success, Support, and Operations. Sit in on calls, watch how people actually work, and find where time and accuracy are being lost.
• Decide what's worth building. Size the opportunity, weigh effort against impact, and be willing to argue that something on the roadmap shouldn't be built.
• Write clear specs and requirements that engineering, support, and leadership can all work from.
• Define what success looks like before you build, in numbers. Track it after launch and be honest about whether it worked.
• Contribute to roadmap and prioritization discussions with a point of view backed by evidence.
• Navigate the tradeoffs that come with a regulated product, where compliance requirements,client-specific rules, and PMS integration constraints all limit what a clean solution can look like.

Building

• Design, build, test, and ship AI-powered features and workflows. This includes prompt design, tool and function definitions, evaluation, and the surrounding application logic.
• Build evaluation sets and test cases so we know whether a solution actually works before it touches a client. Real accuracy numbers, not vibes.
• Prototype fast and iterate in front of real users. Get a rough version working in days, not a perfect version in months.
• Partner with engineering on anything that needs to move into the production platform, and document your work so it can be maintained by others.
• Monitor what you ship. Track accuracy, failure modes, cost, and adoption after launch, and improve it.

What we're looking for:

Required

• 3 or more years in a product role: product manager, technical product owner, product operations, solutions or forward-deployed engineer, or founder. You've owned features or products end to end, not just executed a backlog someone else wrote.
• A track record of shipped work you can point to and explain: what problem it solved, what you decided not to build, how you knew it worked.
• Strong product judgment. You can look at a workflow, iden