Requirements

• Experience building systems software in networking, storage, runtimes, databases, distributed computing, or inference
• Deep understanding of Linux processes, scheduling, virtual memory, I/O, networking, and isolation
• Ability to measure performance before optimizing, using profiling and tracing tools
• Strong reasoning about concurrency, recovery, resource contention, and balancing performance, reliability, and complexity
• Technical curiosity to independently investigate unfamiliar systems and understand their behavior
• Clear communication skills and ownership to explain tradeoffs, make uncertainty visible, ship changes, and operate them in production
• Experience is particularly relevant with eBPF, RDMA, confidential computing, accelerator infrastructure, or open-source systems projects
• At least three to six years of full-time experience working on production systems

What you'll be doing

• Build foundational software for agent execution, model serving, and shared services with focus on speed, safety, and reliability
• Develop secure runtime and isolation mechanisms using Linux, containers, and microVMs
• Improve model-serving throughput, latency, accelerator utilization, and cost
• Investigate Linux CPU, memory, filesystem, and network behavior to optimize resource scheduling and workload isolation
• Maintain correctness of distributed execution with retries, crashes, and recovery while ensuring tenant isolation
• Enhance storage and data systems for consistency, versioning, movement, and recovery of customer data
• Implement accurate usage metering for model tokens and compute time despite concurrency challenges
• Operate what you build and remain accountable for its behavior in production

Nice to haves

• Experience with eBPF, RDMA, confidential computing, accelerator infrastructure, or systems-level open-source projects
• Background in cloud providers, networking companies, database teams, HPC, or trading infrastructure

Perks and benefits

• Work autonomously in small squads owning their roadmap, technical decisions, and production operation
• Asynchronous communication with clear documentation and design processes before meetings
• Use AI tools for productivity with automated tests and reviews
• Investments in observability, migration tooling, automated recovery, and runbooks
• Ownership of shipped changes including production fixes when needed