Founding Backend Engineer

Proximitty is the AI operating system for commercial lending, automating commercial underwriting and loan servicing for banks and fintechs. The company takes a process that typically runs 150 days down to under 2 weeks. Three loan types: commercial real estate (CRE), commercial & industrial (C&I), and SBA (in active design partnership with a customer). The product is Proximitty Servicing Studio, AI agents that autonomously monitor loan portfolios, contact borrowers for documents, parse any format (including blurry scans and handwritten notes), and spread financials using each institution's exact business logic.

The Agent Studio captures every servicing rule, assumption, and edge case loan officers carry in their heads, then automates them. Every spread is verified by Proximitty's in-house commercial lending experts for 100% accuracy. The market: $3T commercial lending market. It costs banks $25K on average to originate one commercial loan. Proximitty can charge up to a third or half of that per loan, meaningful unit economics on a massive base.

The opportunity is white space: frontier LLMs (Claude, GPT) can't automate commercial underwriting yet because the nuances and assumptions per industry vertical (healthcare, manufacturing, CRE) are too specific. Proximitty is building industry-specific commercial underwriting models as the moat. YC W26 batch. Founded November 2025 in San Francisco. Backed by Y Combinator, CRV, Cohen Circle, Uphonest, with angels leading teams at Stripe and OpenAI.

Proximitty is hiring three Founding Engineers. This person owns product and engineering end-to-end while also traveling roughly once a month around the country to speak directly with customers. The mandate is to take Proximitty's 8 active pilots and make the product solid enough to convert them to 7-figure ARR contracts. The immediate priority is owning one agent end-to-end, the borrower portal experience as the first surface, including computer vision models that parse financial statements and reconcile discrepancies across statements in real time.

If a borrower uploads a bunch of financial statements, the biggest challenge isn't parsing the statement itself, it's the cross-statement discrepancies. That's complicated to build. They are really looking for a superstar, ex-founder, somebody with founding experience, somebody who's super learns quickly, doesn't have to be a domain expert, and somebody who speaks to customers. The hire is the third founder in spirit.