I'm building an AI-native operations tool for private mortgage lenders. MVP targets two workflows that eat servicing staff time: NSF collections follow-up and renewal management. I run a Canadian private lender as my day job, so the product is built against a real shop with real data and real users. Self-funded. No VCs, no board.

The data model is locked (20 tables, multi-tenant by lender_id, idempotent CSV imports, append-only event log). Wireframes in progress. Hiring one engineer for the full ~6-month build via a paid bake-off: $5k flat for ~1 week of work, 2-3 candidates in parallel, winner takes the engagement. Losing candidates get paid in full and keep their work.

Trial scope: CSV loan tape import with idempotent upsert + audit, one case detail screen, one LLM call classifying a borrower email reply with structured output, activity log driven by case events. Stack is contractor's choice — Postgres required, Next.js + TS or Python + React expected.

Looking for senior generalists with real production LLM experience (structured outputs, evals, failure modes — not side-project tier) and strong relational data modeling judgment. Opinions you'll defend in writing.

Email: info@readifinancial.com

with: one paragraph on why this interests you, 1-2 links to shipped work involving a real schema + LLM in production, your rate and availability, and one question about the project. No cover letters.

Search terms: postgres, multi-tenant, LLM, OpenAI, Anthropic, Next.js,

TypeScript, Python, fintech, vertical SaaS, founding engineer, contract, Canada