Member of the Technical Staff - Customer Tools

Andromeda is a market and infrastructure platform to buy, sell, and operate compute. The role involves building the software AI labs use to run their compute: a new product that lets a customer run clusters from multiple providers as one fleet, see what every GPU is doing, and sell idle GPU hours on the Andromeda marketplace. Most of the work is backend in Go against Kubernetes and Postgres, with a TypeScript console on top. This is an IC role.

Responsibilities

Build the software AI labs use to run their compute; Build the first versions of the product with principal engineers and with the research and infrastructure teams at customers; Work mostly on backend in Go against Kubernetes and Postgres, with a TypeScript console on top; Manage each customer's clusters through the backend; Carry a pager and talk to customers every week; One fleet: let customers run clusters from Andromeda, other providers, or their own sites under one login and one way to run jobs; Visibility: give customers health, usage, spend, and alerts for every cluster in one place; Identity and access: build organizations, roles, and single sign-on (SSO) that work across every cluster a customer runs; Idle capacity: let customers see unused GPUs, offer them on the marketplace, and get them back when needed; Decide what to build next based on customer problems, and ship it

Requirements

Impressive technical work you can go deep on, with impact in the world; Multi-tenant systems: built authorization, organizations, and data separation between tenants; Kubernetes fluency: built against the Kubernetes API, not only deployed to it; Telemetry pipelines: built the metrics, logs, or event pipelines behind dashboards and alerts people act on; Backend depth: shipped production services in Go, or have the systems background to ship production Go within your first month; Enough front end to ship: build the front end for each feature you own; Production on-call experience: carried a pager for a system customers depend on, and made it quieter afterward

Nice to have

ML or HPC infrastructure: training clusters, Slurm, or job schedulers, from the operator's or the researcher's side; Cloud consoles or developer platforms: products where the user is an engineer running real workloads; Enterprise identity: SSO, SCIM, and identity provider integrations; Data movement: syncing or replicating large datasets across clusters or regions

Benefits

Competitive pay; meaningful equity; comprehensive benefits for you and your dependents, including healthcare, dental, and vision coverage; 401(k); unlimited PTO