Data Onboarding Engineer — AI-Assisted Data Migration

Description Read the whole post before applying. The filter here is judgment, not tooling. You must send over a video (you can use Loom) telling us why you are the perfect fit for this role. Applications without a video won't be considered. What we do Flow is the AI-first operations platform for manufacturer sales — rep firms and the manufacturers they represent. CRM, quoting, order tracking, commission reconciliation, pricing, product data, analytics. All one platform. The problem Every new customer arrives with a decade of operational data in the worst possible shape. Commission statements from forty different manufacturers, each with its own layout. Pricing files with tiers, customer-specific net pricing, quantity breaks and per-product overrides. CRM exports where the same customer appears six times under five spellings. Product data scattered across spreadsheets and PDF spec sheets. All of it has to land in our system, correctly, before that firm can go live. If a commission rate is wrong, someone gets paid wrong. If a price is wrong, a quote goes out wrong. There is no "close enough." What we've already built We are not asking you to do this by hand. We have AI skills and workflows that read raw statements and price books, work out the structure, and produce upload-ready files and reusable templates. That machinery is real, it's good, and it does most of the mechanical work. What it can't do is the part that actually matters: It doesn't know the context of the upload. Why this file, for this customer, at this stage of onboarding, and what it has to line up with that's already in the system. It doesn't know what the firm wants out of their own data. How they think about their territories, their principals, their splits, what they want to be able to report on a year from now. That comes out of a conversation, not a spreadsheet. It isn't accountable. Somebody has to own the statement "this went in correctly" and mean it. It can't talk to the customer. It can't get on a call, ask the right question, explain what we found in their file, tell them what we need from them, and keep them confident while their business gets moved onto a new system. You do those four things. The AI does the grinding; you supply the context, the judgment, the accountability, and the relationship. And you do your half through AI too. We don't want anyone living in Excel or hand-writing scripts — you need the data fundamentals to know what right looks like, and you need to get there using modern AI tooling. What you'll actually do Own the data side of onboarding for your customers, end to end. From the first "send us what you've got" to the moment they're live and trusting their numbers. Get on calls with the customer. Understand their business, how they run their territories and commissions, and what they want their data to do for them. Ask the questions the files can't answer. Explain what you found and what you need — in their language, not in data language. Steer the AI workflows that build the pricing, commission, product and CRM upload templates. Direct them, review what they produce, correct what's wrong, and make the calls they can't. Run the uploads and prove they're right. Tie back to the source. Reconcile totals against the customer's own numbers. Find the twelve rows that quietly vanished. You are the last checkpoint before it's live. Be the accountable owner. When a customer asks "is my data right?", you're the one who answers, and you've earned the right to say yes. Make onboarding better every month. You'll see the same problems repeatedly — turn them into better templates, better prompts, better questions to ask up front, a faster path to go-live. We want your opinion on how this process should work, not just your execution of it. Who we're looking for 1. You're excellent with customers. You can run a call with a firm's operations manager or owner, earn their trust, ask sharp questions, deliver bad news about their data without losing them, and keep a migration feeling under control. This is at least half the job. 2. You're relentless about correctness. You don't feel finished until the numbers tie, and you get genuinely bothered when something is off by $14. "It's probably fine" isn't in your vocabulary. 3. You have the data background — and you don't work that way anymore. You understand how data actually fits together: relational structures, primary and foreign keys, joins, cardinality, normalization, referential integrity, what makes a record unique and what happens downstream when it isn't. You've earned that understanding somewhere real. But we are not hiring you to sit in Excel building formulas or hand-writing Python scripts. Those are the old way. Your fundamentals are what let you direct the work and know when the output is wrong — not what you spend your day doing. 4. You are AI-native and obsessed with it. This is the part that's non-negotiable. You do data work through AI now, every day, by default. You've built real workflows with it. You know how to decompose a messy problem into something a model can execute reliably, how to spot when it's confidently wrong, and how to build the check that catches it. If your answer to "how would you reconcile these forty statements" starts with opening a spreadsheet, you're not the person for this. 5. You've seen data like this before. Strong preference for anyone who's worked with manufacturer rep, distributor or industrial commission and pricing data — commission statements, price books, distributor net pricing, product catalogs. ERP, accounting or channel-data experience counts. You'll ramp faster and catch what a generalist won't. Working with us Fully remote, long-term, steady volume. You'll work directly with our team — no layers. We ship fast and expect the same responsiveness back. To apply Generic proposals get archived unread. Answer these in your own words: Describe the messiest dataset you've had to normalize. What was wrong with it, what did you do, and how did you verify you got it right? Tell us about a data migration or implementation where you dealt directly with the customer. What did those conversations look like when something went wrong? What's your experience with commission statements, pricing files, product catalogs, or channel/distributor data? How has AI changed the way you do data work in the last year? Be specific — which tools, what you've built with them, and what you stopped doing by hand.