Senior Software Engineer

About Owlue Americans overpay tens of billions of dollars in property tax every year, and almost nobody fights back. 74% of homeowners have never appealed their assessment. Only about 8% of the 9.3M seniors eligible for a property tax exemption ever claim one. Commercial values have fallen sharply since 2022 while assessment rolls lag behind. The reason the gap persists is not that valuing a house is hard. It is that recovering the money means doing jurisdiction-specific legal work across 3,000+ counties, each with its own forms, deadlines, evidence standards and filing rules. That work has never been cheap enough to do at scale. We built the engine that makes it cheap. Owlue runs property tax appeals end to end: from finding the over-assessed property, through valuation and evidence, to filing and following through with the board. We are live across multiple states, filing at volume this season, and we are paid only when the owner wins, which means our software has to actually be right. We are seed stage, in revenue, and growing quickly. Why this is an unusually good engineering problem The feedback loop is real. Every appeal has a ground truth: the board either grants the reduction or it does not. Every valuation has a ground truth: what the house actually sold for. There are no vanity metrics here. When the system gets better, homeowners get money back and we get paid. The adversary is real. Our work product is read by county assessors whose job is to reject it, and by hearing officers who will not accept a number nobody can explain. That constraint shapes everything we build: what we produce has to be accurate, and it has to be defensible to a skeptical human being who is looking for a reason to say no. The scale is real. National property data measured in billions of records, geospatial resolution down to the parcel, statistical validation over enormous sales histories, and an operational pipeline that has to be correct jurisdiction by jurisdiction or the case is thrown out on a technicality. And the AI is load-bearing, not decorative. We do not sell AI. We run on it, because it is the only reason a business like this works at software margins. If you want your models to be the product rather than the demo, this is that job. What you would actually work on Pick a lane or move between them. All of this is live or close to it. Agent systems that carry real work from end to end, make judgment calls under uncertainty, and are held to a measurable standard of correctness rather than a vibe check.Applied modeling against hard ground truth, including automated experiment infrastructure that lets improvements be proposed, measured and accepted or rejected on evidence rather than opinion.Turning the rules of thousands of jurisdictions into software: documents in, structured and checkable rules out, filed correctly and on time.A data platform at national scale: ingestion, geospatial resolution, entity matching, and the query layer every product sits on.The product surface. Consumer-facing flows, internal operator tooling, and analytics, built in Next.js and React. Stack TypeScript, Next.js App Router and React on the front end. Python for agent and modeling work. Postgres with PostGIS, plus a large-scale analytical data layer. Serverless web plus containerized background workers. Frontier LLMs from more than one provider, used with the guardrails that production requires. We care far more about how you think than which of these you have already used! You might be a fit if You have 7+ years building and shipping production software, with real depth somewhere: distributed backends, ML or statistical modeling, data engineering, or LLM systems.You have taken an LLM feature from "demo that impressed everyone" to "runs unattended, fails safely, and we know its error rate." That gap is most of this job.You can own a domain end to end: schema, service, model, and the interface that exposes it.You have opinions about evaluation. You measure before you claim.You can hold real-world messiness without flinching. Public records contradict each other, counties change the rules mid-season, and none of that excuses a wrong number going out the door.Helpful but not required: real estate, appraisal, tax, govtech, geospatial, or document-extraction experience. How we work Small and very senior team, no ceremony, direct access to the founders, and you ship to production in your first week. We are genuinely AI-native internally, not as a talking point. Engineers here run agentic coding tools as part of the normal workflow, review each other's code with a second model before merge, and automate their own toil. If that is already how you like to work, you will have an unusual amount of leverage here. The team you would join includes a CTO who ran Google Cloud's GPU platforms for AI training and serving, originated Android Desktop, was a tech lead on AWS Nitro and co-designed Intel SGX, with 25+ patents; a CEO who has co-founded three startups with an exit; a CRO who has taken a company past $1B in sales; and a Chief Tax Counsel who is a former Chairman of The Appraisal Foundation, which is to say the person who helped write the standards our work is judged against. Compensation and upside Base salary benchmarked to top of market for senior and staff engineers. We do not lowball and then talk about equity. We pay well, and then we talk about equity.Meaningful early-employee equity, not a rounding error. Before you have to decide anything, we will walk you through the cap table, the current valuation, and an illustrative model of what your grant is worth across several outcomes, in writing.Health, dental and vision coverage. Flexible PTO. Seed stage, in revenue, and going after a multi-billion-dollar market with a small team. The equity is the point. Joining now is the difference between being an early engineer and being the person who built it. How to apply Apply here with a resume or a link to something you have built. In one paragraph, tell us about a system you shipped that had to be right, and how you knew it was. Owlue is an equal opportunity employer. We evaluate every applicant on merit and do not discriminate on the basis of race, color, religion, sex, sexual orientation, gender identity, national origin, disability, veteran status, or any other protected characteristic.