Senior Data Platform Engineer

About us: We generate home-improvement leads through paid ads and sell them to contractors (roofing, siding, windows, bath). The business works and the software works. Our data does not: three of our own apps report the same numbers and disagree, and we cannot prove which is right. No number is calculated once and stored as a fact — every screen re-derives it from raw data, every time. What you're walking into: One Supabase Postgres serving live apps AND all reporting (test and production share it too). ~43 hand-built scheduled jobs on Cloudflare Workers. Feeds from ~10 systems: Meta/Google/TikTok ads, Stripe, QuickBooks, GoHighLevel, Airtable — Airtable stays system of record for part of the business for a while. Modest volume (low millions of rows, not petabytes). A CTO who will navigate it with you, and a plain-language metric-definitions cleanup already underway on our side. The engagement: A paid audit, about a week. You study what we have and deliver a written recommendation and scoped plan — "here is a smaller thing you should do instead" is an acceptable conclusion and worth the money. If the plan is right and we work well together, we continue into the build. Neither side commits to that in advance. Outcomes we want (how you get there is what we're hiring you for): 1. A view on how our operational Postgres should actually be laid out — what is kept exactly as it arrived vs. what represents current state. 2. A clean line between what our live apps need from the database and what reporting needs, with a reporting layer on the right side of that line. 3. Every number on any screen traceable to one written definition a non-technical person can read. 4. Any past period recomputable, producing the same answer it produced before. (Today some tables are overwritten on a schedule and that history is gone.) 5. A broken or missing feed that stops the update instead of quietly publishing a wrong number. 6. Something SCALABLE our own team can run and build upon. We intend to 10x in size over the next 5 years and don't want to re-haul our infrastructure twice. Requirements: You have designed this before, more than once, and can point to it. You are willing to be opinionated about the database we already have, not only the layer above it — the two get designed together. Fluent English, some 9–5 EST overlap, comfortable moving fast. Probably not a fit if: You want to rebuild our applications (we like them — this is about the data underneath). You treat the existing database as someone else's problem. You work in one vendor's ecosystem only. You can't explain a design decision to a non-technical person — half this job is getting our team to agree what a number means. To apply, please answer this question: What is the most complex data platform you have personally designed and built end to end — data sources in, modeled warehouse, reporting on top? Please tell us about it. We review proposals fast and will reach out to schedule a call. P.S attached to this job posting is a document that goes more in depth on our existing data architecture and the issues we are facing.