Ad-Spy Platform (Meta/TikTok Ads, Shopify Sales Estimates): Engineer Who Has Scraped This Before

We are building an ad-spy platform for e-commerce sellers: the kind of tool media buyers and dropshippers pay for every month. Users search a library of Meta and TikTok ads, see how long each ad has been running and how fast it grows, open the Shopify store and the product behind it, and get estimated units sold and revenue per product, price history, a daily Top 10 of winning ads and products per market, and image search for the same product across shops. French-speaking market first, then English and Spanish. The design (38 Figma screens), the data model and the brief are finished. WHO WE ARE LOOKING FOR One engineer who has ALREADY collected this kind of data in production and derived numbers from it. You must have done at least two of these yourself, not read about them: - Scraped the Meta Ad Library at scale (your own collector or a published actor): you know how the result cap, the start-date vs delivery-date problem, EU vs non-EU data and blocks actually behave. - Scraped TikTok ads (Creative Center, Commercial Content Library or equivalent) and know what each source gives and does not give. - Crawled Shopify stores and obtained stock levels per variant (products.json does not give them), then turned daily stock changes into units sold and revenue, handling restocks, hidden inventory and failed crawls. - Built any metric no API gives (sales estimates, revenue, ad longevity, spend estimates) and validated it against reality. NOT for generalists. Knowing Next.js, Node and Postgres is the baseline, not the skill. If your experience is full-stack apps, CRUD APIs or UI from Figma, please do not apply: the screens are the easy part and they are already designed. THE WORK (fixed $2,500, 10 weeks, ~40 h/week) 1. Weeks 1-2: schema with append-only daily snapshots, Meta + TikTok ingestion (Apify actors behind our own interface, or your collector), Shopify crawler with stock per variant, S3 media, auth, staging - $500 2. Weeks 3-4: ads library (27 filters, fast on millions of snapshot rows), ad detail, landing-page to product resolution, AI analysis - $600 3. Weeks 5-6: products and shops with sales/revenue estimates, price history, product clustering across shops, daily Top 10 - $600 4. Weeks 7-8: image search (pgvector), suppliers, tracked shops + alerts, AI agent + MCP - $500 5. Weeks 9-10: TikTok UI, Stripe, admin with provider costs and job health, load test with real users, production - $300 Stack: TypeScript/Node, Next.js, PostgreSQL 16 + pgvector, Redis + BullMQ, Docker on a Hetzner VPS. Python-first is fine if you also ship Node. You get the Figma, brief, CLAUDE.md, reference schema, OpenAPI skeleton, seeds, fixtures and acceptance checklists per milestone. HOW WE CHOOSE By what you built and how you answer, not by badges. No Upwork history is fine. A GitHub repo, a published actor or a live product, plus precise answers to the questions below, is what gets you to the next step. Vague or generic answers (the kind an AI writes without having done it) end the conversation. PROCESS: questions -> short written exchange about availability and motivation -> paid 2-hour technical test (crawler + snapshots + sales from inventory) -> contract. Individuals only, no agencies. Start your proposal with the word "inventory" so we know you read this.