Low-Latency AI Inference Engineer

OpenAI; California (MO); On-site; USD 180,000 - 240,000; Full time

Job summary

OpenAI's Inference team builds high-volume, low-latency production and research systems around our largest AI models. This role focuses on optimizing performance, resource usage, and reliability in a production environment.

You will partner with researchers and engineers to deploy cutting-edge techniques, improve latency and throughput, and own end-to-end solutions from design to deployment, including tuning on Azure GPUs, distributed stacks, and monitoring tooling.

Qualifications

• Have at least 5 years of professional software engineering experience.
• Have or can quickly gain familiarity with PyTorch, NVidia GPUs and the software stacks that optimize them (e.g. NCCL, CUDA), as well as HPC technologies such as InfiniBand, MPI, NVLink, etc.
• Experience architecting, building, observing, and debugging production distributed systems.
• Bonus point if worked on performance-critical distributed systems.
• Have needed to rebuild or substantially refactor production systems several times over due to rapidly increasing scale.
• Are self-directed and enjoy figuring out the most important problem to work on.
• Have a humble attitude, an eagerness to help your colleagues, and a desire to do whatever it takes to make the team succeed.

Responsibilities

• Work alongside machine learning researchers, engineers, and product managers to bring our latest technologies into production.
• Work alongside researchers to enable advanced research through awesome engineering.
• Introduce new techniques, tools, and architecture that improve the performance, latency, throughput, and efficiency of our model inference stack.
• Build tools to give us visibility into our bottlenecks and sources of instability and then design and implement solutions to address the highest priority issues.
• Optimize our code and fleet of Azure VMs to utilize every FLOP and every GB of GPU RAM of our hardware.

Skills

End-to-end ownership

Self-directed

Software engineering

Distributed systems

Tools

PyTorch

CUDA

NCCL

MPI

InfiniBand

NVLink

NVIDIA GPUs

Job description

OpenAI's Inference team builds high-volume, low-latency production and research systems around our largest AI models. This role focuses on optimizing performance, resource usage, and reliability in a production environment.

You will partner with researchers and engineers to deploy cutting-edge techniques, improve latency and throughput, and own end-to-end solutions from design to deployment, including tuning on Azure GPUs, distributed stacks, and monitoring tooling.