Engineering · Remote (US) · Full-time (future)
Backend & Infrastructure Engineer
Own the cloud, the agent orchestration platform, and the systems that keep claims flowing reliably.
About the role
You'll own ClaimIt's backend and cloud infrastructure end-to-end — from the FastAPI gateway that fronts our agents, to the Pub/Sub-based async orchestration, to the MongoDB data layer, to the Cloud Run deploys and IAM. We're built on GCP with a small, opinionated stack: we want fewer moving parts, not more. You'll set the patterns that the rest of the team follows.
What you'll do
- Own the api-gateway service (FastAPI, Python) — design new endpoints, debug performance, evolve the data model
- Run the agent orchestration platform: Pub/Sub topics, Cloud Run workers, retries, observability
- Architect and operate the MongoDB layer (schemas, indexes, migrations, query patterns)
- Manage GCP infrastructure via Terraform (Cloud Run, IAM, Storage, Pub/Sub) — including disaster recovery and cost optimization
- Lead engineering hires on the backend side as we grow
What you'll bring
- 4+ years of production backend engineering, ideally in Python (FastAPI or similar async frameworks)
- Real cloud infrastructure experience on GCP or AWS (Cloud Run / Lambda, IAM, storage, pub/sub-style messaging)
- Comfort with Terraform or equivalent IaC
- Strong opinions about reliability, observability, and graceful degradation
- Comfort owning an entire service stack — you debug what breaks, regardless of layer
Nice to have
- Experience with LLM serving infrastructure (vLLM, streaming, multi-tenant inference)
- Background in regulated or high-trust domains (finance, healthcare)
- Past work as the first or second backend hire at a startup
Submit your interest
We'll reach out when we open this role.
Other roles: Frontend Engineer·Agent Logic Engineer
About ClaimIt
ClaimIt is building practical agents for tedious consumer workflows — starting with post-purchase price drops. We're early, technically opinionated, and focused on shipping useful systems rather than benchmarks.