SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Salary: USD 278,530 - 345,040 / annual
Roblox is seeking a Principal Software Engineer to lead the Agent Infrastructure team, which builds the systems powering autonomous AI agents for millions of creators and thousands of internal engineers. The team solves unique challenges around running long-duration, non-deterministic agent workloads safely and cost-effectively at scale—a problem space that barely existed two years ago.
In this role, you will own the design and implementation of Roblox's core agent infrastructure, scaling it to serve both internal and external customers. You'll drive reliability and security by keeping long-running agents alive through deploys and outages, implementing sandboxing to isolate untrusted code and control credential access, and defining SLOs and observability standards. You'll partner across Security, Engineering Efficiency, and Infrastructure teams to align the agent platform with broader systems. As an early member of this team, you'll mentor senior engineers, lead design reviews, and push the boundaries of what agents can accomplish—including using agents to build the agent platform itself.
You'll also stay at the forefront of agentic workflows, experimenting with swarms, loops, and orchestration patterns, and bringing validated approaches to the broader Foundation AI organization.
Roblox operates on a hybrid schedule: Tuesday, Wednesday, and Thursday in-office at San Mateo headquarters, with optional presence on Monday and Friday.
REQUIREMENTS:
- 8+ years building and operating backend, distributed, or infrastructure systems in production, with systems you designed that held up at scale for years
- Deep expertise in at least one core domain: workflow orchestration, scalable backend infrastructure, agentic runtimes, or secure sandboxing and access
- Hands-on builder who spends most time writing production code in Go, Python, Rust, or similar languages
- Agent-native: active experimentation with swarms, loops, and orchestration; strong opinions on harness and model choice; constantly evolving agentic workflows
- Proven experience running agents against real systems in production, understanding failure modes and infrastructure requirements
- Bachelor's degree or equivalent experience in Computer Science, Computer Engineering, or similar technical field
NICE TO HAVE:
- Security experience with untrusted code execution: sandbox and container isolation (gVisor, Firecracker, microVMs), Kubernetes hardening, network segmentation and egress control
- Experience with Temporal, AWS, and self-hosted models
- Production Kubernetes deployment experience using GitOps and Argo CD
- Open source contributions or experience building and shipping a product with real user traction
