Staff Software Engineer, Agentic Infrastructure

Location: Long Beach, California, United States

Department: G-Dark Matter Lab

About the Team:

Dark Matter Lab is a research group within Relativity Space focused on advanced aerospace systems, agentic engineering, and technologies outside the conventional roadmap. We build the infrastructure needed to turn new ideas into engineering capabilities people can actually depend on.

About the Role:

We run a stack of interconnected agentic systems — SybilClaw, yapCAD, Mechatron, Multigraph, local LLM inference, an inter-agent message bus, parametric CAD pipelines, a print farm, and the infrastructure connecting it all. We need an engineer who can own and evolve these systems as they move from prototypes into production engineering infrastructure.

This role is for someone who has deployed an agentic harness (OpenClaw, Hermes, SybilClaw, or something comparable) in a real work environment. You understand what it takes to make these systems reliable when engineers depend on them every day. You’ll learn the stack, improve it, and work with forward-deployed engineers to package portions of it for deployment with internal and external customers.

• Own the reliability of our agentic infrastructure: agents, sessions, model routing, context pipelines, inter-agent communication, and supporting services
• Build and maintain infrastructure where good off-the-shelf solutions don't yet exist
• Deploy and operate local LLM inference across Mac Studio and GPU hardware
• Manage Linux/macOS systems, Proxmox VMs and containers, storage, backups, and recovery
• Maintain multi-site networking including L3 routing, VLANs, DNS, firewalls, and connectivity
• Build CLIs, dashboards, automation, and internal tools that make engineers faster
• Improve observability, debugging, and automated failure recovery
• Contribute upstream to open-source projects we rely on
• Help design practical security around local compute, data handling, and access controls
• Package and deploy portions of the stack into internal and customer environments
• Work directly with engineers to understand what they need and turn recurring problems into better infrastructure

About You:

• 7+ years of experience building and operating complex software and compute infrastructure, with hands-on work across hardware, operating systems, networking, and automation
• Experience deploying an agentic harness such as OpenClaw, Hermes, SybilClaw, or a comparable system in a real work environment
• Strong Linux and macOS systems experience, including debugging services, processes, storage, permissions, and networking
• Experience operating physical infrastructure, VMs, or containers in production
• Strong networking fundamentals including routing, VLANs, DNS, and firewalls
• Ability to build software, scripts, CLIs, and integrations when existing tools aren't sufficient
• Meaningful experience contributing to or maintaining open-source software
• Strong judgment around reliability, security, performance, and simplicity

Nice to haves but not required:

• Experience with ITAR-regulated, air-gapped, export-controlled, or similarly constrained environments
• Experience operating local LLM inference with Ollama, MLX, vLLM, llama.cpp, or similar
• Experience with model routing, context management, tool execution, agent state, or multi-agent systems
• Experience building internal developer infrastructure used daily by other engineers
• Experience deploying systems into customer or forward-deployed environments
• A GitHub, Gitea, or other body of work where we can see what you've built

This role requires in-office presence at least three days per week, with flexibility to work remotely when the work allows. Much of this infrastructure is physical, local-first, and best built alongside the engineers using it.