Senior Full Stack Engineer (AI)

A client's rental properties are often 30 to 50% of their net worth. The tool their financial advisor uses to think about them is a spreadsheet.

Leveridge takes a client's tax return and turns it into every property they own in one view, then models what happens if they hold, sell, or exchange. The advisor makes the call. We do the math.

We launched in mid-May. 13 advisory firms pay for Leveridge, up from 7 in early August.

We're four people, two founders, our founding engineer, and a front-end engineer. You'd be the fifth, and the third engineer, which means you'd touch nearly all of it and your name would be on a lot of it. We're in the middle of raising a pre-seed.

What you'd own

An advisor sits down with a client and has to say what to do with the largest asset that client owns. Hold it, sell it, or exchange it. The numbers on the screen when they make that call came from our code, and they're going to defend those numbers out loud with their license behind them. That's the job.

Today, when an advisor hits a hard case, they email us, or ask for fifteen minutes before the client meeting, or want to walk back through it afterward. What they're pulling on is what we know about real estate, and too much of that still lives in our heads instead of in the product. Getting it into the software is the work. The advisor should walk in alone and be the expert in the room.

The math has to be exactly right, and the data it runs on doesn't exist anywhere yet. Purchase price is in a closing file. Basis and depreciation are on a tax return, often a different one each year. The loan is on a lender's statement, the value is somebody's estimate, and part of it is in no system at all and has to be asked of the owner, because the owner is the only one who knows. A household arrives across several documents and several years, the same property reads differently in each, and the system has to reconcile that instead of quietly averaging it. If the data is wrong, the recommendation is wrong, so nothing downstream runs on a number nobody has stood behind.

Underneath that the engine is deterministic. Depreciation, basis, recapture, federal and state brackets, to the dollar. AI reads the documents and writes the narrative. It doesn't touch the math.

Portfolios have had a system of record for decades. Real estate never has.

You'd start on the document pipeline, which turns tax returns into the data everything else runs on. It works today, it isn't good enough, and nobody has owned it end to end. After that the roadmap is mostly open and you'd have a real say in it. A conversational layer over a client's whole portfolio. Monitoring that tells an advisor something changed before the client calls them. None of it is built.

None of that is every week, though. It's four people, so some weeks it's clearing a stack of bugs because the product needs that more than the roadmap does. That's the job too.

Who we're looking for

You've shipped production software on a real team for four or more years, and you can take an idea from nothing to running in production on your own. That matters more here than anything else on this list.

You've done applied AI work, not prompt wrapping and not research. You've built something where a model sat inside a pipeline that had to be right, and you had to measure whether it was.

A background in fintech, wealthtech, or financial planning helps. It isn't required, and neither is real estate. We'll teach you the domain.

The people who do well here bring us a problem while they're still inside it, instead of only showing finished work. They find the edge cases themselves. Interest-only loans, ARMs, HELOCs, a property carrying four paid-off liens. We need you hunting for what breaks. They say when something will be done, and then it's done, or you hear why before you have to ask. At four people, we're each counting on that.

How we work

Standup every morning, and that's most of the process. A decision gets made in a huddle the same day instead of in a planning cycle. You'd work directly with both founders and with Davis, our founding engineer, who spent years building financial planning simulation software before this. You'd also meet the people using what you build. We're on calls with advisors every week, and what they tell us lands in the product in days, not quarters.

It's also a lot of work. Four people are building something advisors use with real clients, and every one of us has something on the line. Everyone here works hard, founders included, and we hold each other to a high bar. If that's the team you want to be on, we should talk.

Comp

You pick where you sit on the range, more cash and less equity or the other way around. It's below market. What you get instead is a real piece of the company, a surface you own outright, and a say in what gets built. Neither founder has taken a salary yet, so it's the same bet we're making.

How we hire

We read every application. If it looks like a fit, you'll have a short call with me or my co-founder about how we work. Then a working session with Davis on a real problem, no whiteboard algorithms. If that goes well, we make an offer.

To apply

Skip the cover letter. Start your message with one sentence on the last thing you shipped that you're proud of, and why it was hard. Then tell us what you'd want to own here.

Skip the line

Want to jump in front of all the other applications? Send a GitHub link to something you built that shows your skills and the quality of what you ship. If you don't have any public projects, send a GitHub link to this one: create a simulation engine where you simulate buying and selling real estate over a period of time, and include some form of visualization. It's intentionally vague, so show us some creativity.