AI Accelerator Runtime Systems Engineer

Openai; United States; On-site; USD 180,000 - 240,000; Full time

Job summary

OpenAI's Hardware team is building the runtime that turns compiled programs into efficient execution on our AI accelerator. You will design and implement the low-level device runtime, manage memory and address spaces, and coordinate synchronization across domains.

You will work across compiler, kernel, architecture, and silicon teams to define interfaces, validate behavior with simulators, and diagnose hard performance issues.

Qualifications

• Strong low-level systems programming in C, C++, or Rust.
• Experience building runtimes, drivers, firmware, or accelerator software.
• Deep understanding of concurrency, memory ordering, and DMA.
• Familiarity with event-based, cycle-accurate simulators and performance models.
• Ability to reason about scheduling, latency, throughput, and resource tradeoffs.

Responsibilities

• Design and implement the low-level device runtime for OpenAI custom silicon.
• Build kernel-launch scheduling, command submission, queueing, and dependencies.
• Manage device memory spaces, DMA, address spaces, and data movement.
• Implement synchronization primitives, events, barriers, streams, and ordering.
• Define interfaces between runtime, drivers, firmware, kernels, and higher-level systems.
• Use cycle-accurate simulators to develop and validate runtime behavior.
• Diagnose concurrency, memory-ordering, deadlock, race, and performance issues.
• Build tests, tracing, profiling, observability, and reproducible workloads.
• Partner with architecture and silicon teams to improve HW-SW interfaces.

Skills

Low-level systems programming

Runtimes/drivers/firmware

Concurrency and memory-ordering

Memory management and hardware interf?

Event-based simulators

Scheduling and performance reasoning

Debugging across HW/SW

Cross-disciplinary collaboration

Job description

OpenAI's Hardware team is building the runtime that turns compiled programs into efficient execution on our AI accelerator. You will design and implement the low-level device runtime, manage memory and address spaces, and coordinate synchronization across domains.

You will work across compiler, kernel, architecture, and silicon teams to define interfaces, validate behavior with simulators, and diagnose hard performance issues.