Backend Software Engineer

AI Platform scales SPREEAI's infrastructure: productionizing ML model checkpoints, running the API services behind the Partner Portal and partner integrations, optimizing inference serving, and giving ML Scientists training-as-a-service so they can iterate without managing infrastructure themselves. It's the layer between research and the product our partners actually touch.

As SPREEAI's partner roster and product surface area grow, the API services powering partner-facing features and internal ML tooling need dedicated backend ownership instead of being split across whoever has time. These two hires own backend services end to end, from design through production, working alongside AI Platform and Partner Engineering.

• →Design, build, and operate backend services in Go or Node.js: REST APIs, auth, and the data layer behind partner-facing and internal tooling
• →Own webhook systems for partner integrations (delivery guarantees, HMAC signatures, retry logic)
• →Design API versioning, pagination, and idempotency so existing integrations don't break as we ship
• →Work with PostgreSQL and DynamoDB, and make real calls on schema and query design under production load
• →Partner with AI Platform on the services that productionize ML checkpoints and serve inference, and with Partner Engineering on partner-facing API contracts
• →Help set backend engineering practices, code review norms, and service patterns as the team scales

• 3+ years building and operating production backend services in Go or Node.js
• Real experience with REST API design (versioning, pagination, idempotency keys) and auth (OAuth 2.0, API key management, JWT)
• Experience building reliable webhook delivery (at-least-once semantics, signature verification, retries)
• Comfort with PostgreSQL or DynamoDB query and schema design
• Comfort working independently in an ambiguous, fast-moving startup environment

Experience supporting ML/inference-serving infrastructure or working alongside an ML platform team. Docker and Kubernetes exposure.

You'd rather own a service end to end than hand off a spec, and you think about failure modes, a flaky partner webhook, a key rotation, a bad migration, before they happen, not after. You communicate proactively when something's at risk.