Role Overview

Join OpenAI's groundbreaking Sora team as a Distributed Training Engineer in San Francisco, California. The Sora team is at the forefront of revolutionizing video generation through advanced foundation models. As a hybrid research and product team, we push the boundaries of AI capabilities while prioritizing reliability, safety, and real-world deployment. In this critical role, you'll optimize training throughput for our internal frameworks, enabling researchers to experiment boldly with video models that could transform industries like entertainment, education, and beyond.

This position demands a rare blend of systems engineering prowess, ML expertise, and an unrelenting passion for performance optimization. You'll dive deep into supercomputer architectures, craft bug-free ML code, and collaborate directly with world-class researchers. Based in San Francisco with a hybrid model (3 days in-office weekly), OpenAI offers relocation support to bring top talent to our innovative hub. If you thrive on eliminating inefficiencies and scaling AI to unprecedented levels, this is your opportunity to shape the future of generative video AI.

OpenAI's mission to ensure AGI benefits humanity starts with teams like Sora, where engineering excellence meets cutting-edge research. Contribute to models that generate coherent, high-fidelity videos from text prompts, powering applications from creative tools to scientific simulations.

Key Responsibilities

• Partner with researchers to architect systems-efficient video models, focusing on distributed training scalability.
• Implement state-of-the-art techniques in OpenAI's internal training framework to maximize hardware utilization.
• Conduct in-depth profiling of training pipelines, identifying and resolving throughput bottlenecks.
• Optimize training kernels for GPU/TPU clusters, achieving impressive efficiency gains on supercomputers.
• Develop robust Python codebases that enable seamless experimentation with novel video architectures.
• Analyze and stabilize training dynamics to ensure reliable convergence in multi-modal models.
• Debug complex distributed systems issues in large-scale ML runs, maintaining zero-bug standards.
• Collaborate on deploying video models safely into production environments for widespread impact.
• Monitor supercomputer performance metrics and propose hardware-software co-optimizations.
• Document optimizations and mentor junior engineers on best practices in ML systems.
• Stay abreast of advancements in parallel computing, JAX/PyTorch ecosystems, and AI hardware.
• Contribute to safety features that mitigate risks in generative video training pipelines.
• Participate in cross-team initiatives to integrate Sora capabilities into OpenAI's broader product suite.
• Drive continuous improvements in training framework maintainability and researcher productivity.

Qualifications

To excel as a Distributed Training Engineer on the Sora team, you'll need:

• Extensive experience with multi-modal ML pipelines, particularly video and text-to-video systems.
• Deep systems knowledge for optimizing distributed training at supercomputer scale.
• Expert-level Python proficiency, with a portfolio of production ML code.
• Proven success in kernel optimization and performance tuning for AI workloads.
• A passion for dissecting training dynamics and ensuring numerical stability.
• Strong collaboration skills with research teams translating ideas to efficient implementations.
• Familiarity with frameworks like PyTorch Distributed, JAX, or custom training stacks.
• Experience with GPU cluster management, profiling tools (e.g., NVIDIA Nsight, PyTorch Profiler).
• Bachelor's/Master's/PhD in CS, EE, or equivalent, plus 3+ years in ML systems engineering.
• Relentless focus on code quality, performance, and scalability.

Salary & Benefits

Compensation at OpenAI is competitive and includes base salary, equity, and comprehensive benefits. Estimated total compensation for this senior role ranges from $250,000 to $450,000 USD annually, depending on experience. Benefits include health coverage, 401(k) matching, unlimited PTO, parental leave, relocation assistance, and access to state-of-the-art compute resources.

• Full medical, dental, vision insurance
• Hybrid SF work model with catered meals
• Professional development stipend
• Generous equity package in OpenAI
• Wellness and mental health support
• Commuter and fitness reimbursements

Why Join OpenAI?

OpenAI is dedicated to safe AGI development for humanity's benefit. The Sora team exemplifies this by expanding video AI capabilities responsibly. Work with pioneers in generative models, access unparalleled compute, and impact billions. Our San Francisco office fosters collaboration in a vibrant AI ecosystem. We value diverse perspectives and offer equal opportunity employment.

How to Apply

Ready to optimize the future of video AI? Submit your resume, GitHub/portfolio, and a note on your favorite optimization project. We're excited to review applications from passionate engineers.