Applied Ai Engineer

## About Doorvest Doorvest is a vertically integrated, AI-native property management company. We operate across 6 markets and manage 2,000+ doors today, and we're now scaling through M&A by acquiring local property managers and integrating them onto our AI-native operating system in under 60 days. The thesis is straightforward. Property management has historically been a low-margin, labor-intensive business where every new door added headcount and degraded service. AI changes that math, and we're using AI to build the largest residential PM brand in the country. The U.S. market is roughly 240K operators and 10M units with less than 1% share for the largest player. Our team has experience executing on 25+ M&A transactions combined. We're backed by Mucker, M13, Tacora, and founders and executives from Invitation Homes, Wealthfront, Opendoor, and other top fintechs. ## The role A lot of AI work in proptech is a chatbot bolted onto a CRM. We're building something else. Every property management workflow we absorb (leasing, renewals, rent collection, maintenance, homeowner communications, financial ops) gets rebuilt as an agentic system that runs in the background, with humans involved only where the stakes require it. That operating system is how we plan to integrate every PM company we acquire. We're hiring a Founding Applied AI Engineer to own this layer. You'll build the agent infrastructure, the evals that keep it honest, and the product surfaces our operators and homeowners actually use. You'll work directly with the CTO and Head of Product, and you'll set the bar for how AI ships at Doorvest. ## What you'll work on A sample of problems you might tackle in your first six months: - Leasing & renewals. A property manager spends hours each week detecting lease expirations, generating renewal offers, and chasing tenant signatures. You build the agent that runs the full lifecycle end-to-end, along with the eval suite that proves it's safe to leave on. - Rent collection. Delinquency detection, automated tenant notices, payment plan negotiation, eviction processing. You build the agentic case management system that opens, works, and closes most of these without an operator touching them. - Homeowner communications. Multi-channel inbound from thousands of homeowners across the PM books we've acquired. You build the classification, triage, and AI-drafted response system that lets one CX person do the work of five. - Financial operations. Billing, fee processing, utility billback. You build the pipeline that turns these from manual line-items into a fully automated flow. - Integration velocity. Every new PM acquisition arrives with a different stack, different data, and a lot of tribal knowledge. You build the agent-driven onboarding system that compresses integration from quarters to weeks. This is what makes the rollup work. - The platform underneath all of this. Evals, observability, guardrails, and feedback loops. The infrastructure that lets us safely run agents against real homes, tenants, and dollars. ## What success looks like 6 months in - The AI workflows you build are running daily across our acquired PM books. Not pilots, not demos. The ops team would notice if it went down. - A workflow that used to take hours per door per month now runs as a background agent, and doors per ops headcount has measurably moved. - You can point to specific margin expansion (gross margin per door, integration time on the latest acquisition, response time on homeowner inbound) that moved because of work you led. - The agent and eval infrastructure you built is what every new acquisition gets layered onto, and what every new AI feature at Doorvest is built on. ## Tech stack - Frontend: Next.js, React, TypeScript, Material UI - Backend: Node.js, TypeScript, SST, AWS Lambda - Data: Supabase, DynamoDB, Aurora - AI: We use the frontier model APIs directly, and we expect you to bring strong opinions on agent frameworks, eval tooling, and the rest of the stack - Dev: Claude Code, Cursor, GitHub Copilot. AI-native development is the norm here. Deep expertise in any one slice of this stack matters less than the ability to ship reliable AI systems end-to-end. ## Who we're looking for - You've shipped LLM-powered systems in production. Not prompts in a notebook, but actual agents, retrieval systems, or extraction pipelines that real users depend on, with the evals and guardrails to match. - You think in evals. You know that "it works on the demo" is the start of the work, not the end. You've built the harnesses, the labeled sets, and the production telemetry that catch regressions before customers do. - Full-stack range. You can design the agent, ship the UI it lives in, and wire up the integration it depends on. You don't need someone else to finish the loop. - Owner mentality. You see a messy workflow, talk to the operator running it, and ship the thing that fixes it. You don't wait for a spec. - Pragmatic about AI. You know when a fine-tuned classifier beats an agent, when an agent beats a workflow, and when none of it is the right answer yet. - Background. Meaningful recent time spent on applied AI and LLM systems. Real estate, fintech, or vertical SaaS experience is a plus, not a requirement. ## Logistics - Work authorization: Due to current business constraints, we are unable to sponsor or transfer work visas at this time. Applicants must already be authorized to work in the United States. - Hybrid workplace: We're committed to 3 in-office days at our downtown San Francisco HQ. - Compensation: $155K to $185K base, plus meaningful equity. - Benefits: Medical, dental, and vision insurance (with HSA options), open PTO, company-wide team-bonding trips every two quarters, and various physical, mental, and financial wellness perks. If this sounds like the role for you, we'd love to meet you.