Staff Platform Engineer - Dev Infrastructure
Core
Own the build graph, CI capacity, and deployment infrastructure for a mixed C++/Python/Rust/CUDA monorepo supporting humanoid robots in heavy industry.
Role type
Staff Platform Engineer (Dev Infrastructure)
Builds
Production-ready robot software, reproducible build artifacts, and self-service deployment tooling for autonomous robots.
Domain
Robotics / Heavy Industry / DevOps
Deliverable
production ML models | product features | infrastructure
Required skills
Linux systems administration, CI/CD platform ownership, large-scale C/C++ build systems (Bazel/CMake), Infrastructure as Code (Ansible/Terraform), Python, Bash, GitOps, observability stack management, cross-compilation, artifact management.
Preferred skills
Embedded Linux targets (ARM64, Yocto), hybrid on-prem/cloud environments, robotics/aerospace background, Nix, reproducible-build tooling at scale.
Technologies
Ansible, Bazel, Bash, CI/CD, CUDA, GitHub, GitLab, GitOps, Grafana, Jenkins, Kubernetes, Linux, OpenTelemetry, Prometheus, Python, Rust, Terraform, VPN
Responsibilities
Own the build graph for a mixed-language monorepo including incremental correctness and remote caching. Operate CI capacity including self-hosted and GPU runners to maintain sub-10-minute feedback loops. Improve build reproducibility for bit-identical artifacts across time. Manage versioning, artifact promotion, and container registry infrastructure. Design safe, resumable robot rollouts with staged channels and rollback capabilities. Own signing and provenance for deployed robot software. Create reproducible development environments across laptops and robots. Build self-service tooling for autonomy engineers to deploy branches without tickets. Shape the onboarding path for new engineers to build, test, and deploy on day one. Support cloud and on-prem compute, storage, and operational layers for training clusters. Implement fleet observability across metrics, logs, and traces. Support deployment site infrastructure including networking and offline-capable registries.
