ML Ops Engineer

Location: San Francisco

Department: Engineering

Location Type: IN_OFFICE

Employment Type: FULL_TIME

Who We Are

The Role

What You’ll Do

• Own and evolve the deployment lifecycle for our perception systems across edge and cloud environments.
• Design and manage highly available ML serving infrastructure, ensuring high performance, low-latency inference, and reliability in production.
• Build resilient CI/CD pipelines for testing and pushing system updates with confidence and comprehensive fleet observability.
• Implement and manage remote system monitoring, alerting (e.g., Prometheus, Grafana, Sentry), and debugging systems to ensure operational excellence, focusing on fleet health metrics (e.g., uptime, resource utilization, inference latency).
• Work closely with perception and backend teams to design deployable systems that are robust in the real world.
• Integrate and maintain experiment tracking and model management platforms (e.g., Weights & Biases, MLflow) to streamline model lineage, performance comparison, and versioning from research to production.
• Contribute to security policy design and device authentication/attestation infrastructure for fleet safety.
• Build and maintain internal tooling and CLI utilities to streamline the end-to-end development-to-deployment workflow, empowering the broader engineering team to ship perception systems with high velocity and minimal friction.

What You Bring

• 3-5+ years experience in DevOps, deployment engineering, or site reliability, ideally with production ML systems or robotics.
• Deep operational experience with Linux system administration, system packaging (e.g., Deb/RPM), and configuration management tools (e.g., Ansible, SaltStack, Chef).
• Strong experience with ML deployment/serving frameworks and infrastructure (e.g., PyTorch Serve, custom C++ inference services).
• Comfortable working in Linux-heavy environments with advanced shell scripting and strong knowledge of operating system internals.
• Hands-on experience with networking fundamentals, including TCP/IP, firewalls, NAT traversal, and VPNs.
• Prior experience with managing large-scale edge fleets, including over-the-air (OTA) updates and blue-green deployment strategies.
• A proven track record of developing internal developer tools or CLI applications that automate complex infrastructure tasks and improve overall team productivity.

Nice to Have

• Experience deploying AI/ML inference pipelines on bare-metal or virtualized edge hardware (e.g., using GStreamer/Deepstream pipelines, custom executables).
• Expertise in machine learning inference engineering, including quantization and compilation (e.g., using ONNX Runtime, TensorRT), for efficient deployment to various edge hardware targets (e.g., NVIDIA Jetson, custom ARM SoCs).
• Familiarity with writing or debugging high-performance, low-latency ML inference services in C++.
• Exposure to remote logging, log ingestion, and distributed telemetry aggregation.
• Previous experience in early-stage startups or fast-paced hardware/software integration environments.

Why Sauron

• We celebrate as a team and troubleshoot as a team.
• The goal is the mission, not the credit.

• Be ruthless with problems, but kind to people.
• Raise the bar, lower the shield

• Your perspective is a requirement, not a suggestion.
• Speak the hard truths early so we can fix them fast.

• Do what you say you’ll do.
• If it breaks, fix it. If it works, make it better.

• Earn trust through empathy and consistency.
• Anticipate needs before they become requests.