Staff Infrastructure Engineer, Robotic Systems
Atomicmachines · Emeryville, or Santa Clara, California
About this role
## Atomic Machines — Staff Infrastructure Engineer (Robotic Systems) ### About the Role Atomic Machines is building the **Matter Compiler™**, an AI-native, all-digital manufacturing system that builds working machines directly from code. The Matter Compiler targets **micro-machines**—including motors and gears at the grain-of-sand scale, compact robots, and cooling built directly into AI chips. This role owns the infrastructure that powers the fleet of nodes controlling real-world equipment—covering the **build system**, **CI/CD pipeline**, **artifact store and shared cache**, and the **software deployment path** to ~80 nodes today and thousands of devices across multiple metros by 2027. You’ll own the infrastructure end to end: **design, build, operate, and continuously improve**. You’ll set blast-radius limits, define signals for early CI degradation, and drive adoption so the new infrastructure becomes the default. *Hardware experience is not required.* You should be able to decompose ambiguous failures, reason across abstraction layers, and own problems spanning teams and org boundaries. ### What You’ll Do - Own the build system for a **40-package monorepo** targeting **three architectures** (hermetic builds, honest lockfiles, change detection, and a shared remote cache keyed on real inputs). - Own **CI/CD end to end**, including **rollback**. Measure and improve push-to-green, queue depth, and fetch error rate. - Design and operate the **fleet deployment path**: immutable artifacts, staged rollout, version handshake, drift detection, and an abort path not dependent on the deploy script. - Define production pipeline health **signals and thresholds**, and create a runbook on-call engineers can run without you. Convert runbook steps into pipeline checks and automate remediation where it pays back. - Own **build-runner security** as a blast-radius problem (short-lived credentials, scoped IAM, rotation, and secret scanning in CI). - Partner with IT on **GitLab, network, and identity boundaries**—ask where the boundary is instead of assuming. - Reduce toil with measurable outcomes: what engineers did manually before, how often, and what they do manually now. - Drive adoption across teams that don’t report to you—make the new way the default and slow the old way. - Raise reliability standards without blocking delivery; encode incident learnings into systems, not reminders. ### What You’ll Need - Proven ownership of a build system for a multi-package repo with multiple build targets, including diagnosing issues and implementing changes. - Ownership of CI/CD end to end (including rollback), with an opinion on what makes deployments safer over time. - Deep fundamentals in **build reproducibility and caching** (why local passes can fail in CI; what belongs in cache keys; how a correct key can still produce a wrong artifact). - Deep fundamentals in **fleet deployment** (why mutable tags cause drift; deploying by digest; safe rejoining of offline nodes). - Experience designing concrete failure paths (timeouts, safe states left behind, and what the next engineer sees). - A track record diagnosing production incidents to root cause with a mechanism, measurement, and fix. - Evidence of getting other engineers to adopt infrastructure you built, including the lever you used. - Candor about what you’re not good at and ownership of outcomes (including misses). - Fluency with AI tools in daily engineering work while owning the decisions, code, an
Listing freshness
CronJobs last confirmed this listing 1h ago. If its source stops confirming the opening for seven days, this page is removed from active inventory.