Sobre

We run a SaaS that unifies Amazon and Mercado Livre sales, inventory, and logistics data for Brazilian sellers. Today the materialized view and API group metrics only by product/SKU, so every Mercado Livre variation inherits the same total units and revenue, breaking per-variation tracking and date filtering. On the frontend, the marketplace cell hardcodes the U.S. flag and shows a bright red NULL badge whenever fulfillment is missing, while the backend forces 'FULL' or 'FBA' regardless of the real logistics mode. Variation labels come from a single attribute lookup on the last Mercado Livre order, so the dashboard can display colors we never sold, and inventory falls back to zero whenever a variation record isn’t matched precisely. Our existing integrity service only compares database tables against each other, so stale or incorrect marketplace data slips through unnoticed. We need an experienced engineer to restore trustworthy, real-data visibility while keeping the system simple and resilient. What You’ll Do Deliver per-variation sales history: Extend the ingestion and reporting pipeline so the materialized view and /api/sales-unified/summary return distinct units/revenue per seller_sku + variation_id, while preserving the current filters and pagination. Feed the warehouse with real orders: Ensure Amazon SP-API and Mercado Livre syncs persist dated line items into unified_sales_lines, enabling accurate day filters and avoiding reliance on placeholders or mock data. Fix marketplace presentation: Pass the correct country and fulfillment props to MarketplaceFlag/FulfillmentBadge so Brazilian listings show the right flag and never render the literal string “NULL.” Surface real logistics modes: Replace the hardcoded 'FULL'/'FBA' logic with values pulled from Mercado Livre logistics fields and Amazon fulfillment channels, mapping them to FLEX/FULL/FBA/DBA as returned by each API. Show accurate variation metadata: Swap the “last order” lookup for Mercado Livre catalog attributes so color/size names always match what’s active in the listing. Correct live inventory counts: Update the inventory join so we sum valid variation quantities instead of defaulting to zero when a single row doesn’t match. Build a real-data validator: Extend the existing integrity service to fetch fresh payloads from Amazon and Mercado Livre, comparing them with dashboard data and flagging discrepancies before users notice. Required Expertise Proven experience with Node.js/Express, TypeScript, and PostgreSQL (views, materialized views, query tuning). Familiarity with Amazon SP-API and Mercado Livre REST endpoints, including OAuth token refresh and order/inventory resources. Frontend skills in React/TypeScript/Tailwind for tightening UI components without adding complexity. Comfort designing lightweight data-quality checks and observability hooks in production services. Nice to Have Prior work integrating multi-marketplace seller tools in Latin America. Experience with real-time validation or reconciliation pipelines. Engagement Details Scope: Diagnose, implement, and document the fixes above; no placeholders, mock data, or shortcuts—only real marketplace and Postgres data. Collaboration: Work closely with our founders; confirm API connectivity before each change and keep the solution as simple and robust as possible. Deliverables: Merged PRs with passing checks, concise technical notes, and a repeatable validator that keeps the dashboard honest. How to Apply Please share: Brief summary of similar marketplace integrations you’ve shipped. Approach for backfilling variation-level order history without disrupting production. Example of a data-quality or validation system you implemented.