We own the service that first receives product discovery requests (autocomplete, search, browse & recommendations) & executes the entire business logic that drives product discovery.
That service also functions as a gateway to parse incoming user requests, orchestrate calls across the internal services that do the ranking and retrieval, interpret their responses, and assemble the final response the customer sees.
We also own the public configuration APIs customers use to control how discovery behaves, and the MCP layer that exposes all of it to AI agents.
Our team owns both platform & product components: Because of the positioning of our services, we take part in every architectural decision of our core product offering.
At the same time, our goals are heavily influenced by direct customer feedback, because all product discovery-related features are either implemented in or integrated with our services.
All of our services are public APIs, so interface design and backward compatibility are a constant part of the job.
But most of our attention goes to keeping them fast and up: we serve up to 15k RPS, fan out to many internal services on every request, and are expected to stay available and degrade gracefully when any of them fails or slows down.
The work is distributed systems in practice rather than in theory: request orchestration, failure isolation, caching, latency budgets, capacity planning, and the observability to know what is actually happening in production.
The team is small enough that engineers own problems end to end, from design through on-call.
About youYou're a staff-level engineer who wants to own systems, not ticke.