Role Overview

The Frontier Evals & Environments team at OpenAI is responsible for creating ambitious benchmarks and evaluation frameworks that measure and steer frontier AI models. Our open-sourced evaluations like GDPval, SWE-bench Verified, MLE-bench, PaperBench, and SWE-Lancer have set industry standards. We've conducted frontier evaluations for groundbreaking models including GPT-4o, o1, o3, GPT-4.5, ChatGPT Agent, and GPT-5.

As a Research Engineer, you'll push the boundaries of AI capabilities measurement, owning end-to-end projects that influence training, safety, and launch decisions. This role demands exceptional engineering talent passionate about AGI safety and rapid model progress. Experience firsthand how OpenAI's models evolve and contribute to steering them responsibly.

Working in our dynamic San Francisco office, you'll collaborate with top researchers to build self-improvement loops, RL environments, and scalable evaluation systems. This is your opportunity to make history in AI safety research.

Key Responsibilities

• Create Ambitious RL Environments: Design reinforcement learning environments that rigorously test frontier models' limits and capabilities.
• Measure Model Capabilities: Develop comprehensive frameworks to evaluate skills, behaviors, and emergent abilities in cutting-edge AI systems.
• Innovate Evaluation Methodologies: Pioneer automatic exploration techniques to uncover hidden model behaviors and failure modes.
• Steer Frontier Training: Influence training decisions for OpenAI's largest model runs, gaining early access to breakthrough capabilities.
• Build Scalable Systems: Architect processes supporting continuous, high-throughput model evaluation at scale.
• Implement Self-Improvement Loops: Create automated systems that enhance model understanding and iterative improvement.
• Conduct Red-Teaming: Systematically identify vulnerabilities using creative, adversarial testing approaches.
• Analyze Evaluation Data: Apply statistical rigor to interpret results and inform critical safety decisions.
• Cross-Functional Collaboration: Partner with research, safety, and product teams to operationalize evaluations.
• Open-Source Contributions: Publish frameworks that advance the broader AI safety ecosystem.
• Monitor Production Systems: Implement observability for real-world model deployments.
• Prototype Novel Benchmarks: Rapidly iterate on evaluation environments like SWE-bench and MLE-bench.
• Drive Empirical Research: Lead studies spanning the full spectrum of AI capabilities measurement.

Qualifications

Required:

• Deep passion for AGI/ASI measurement and AI safety research
• Exceptional engineering skills with ML research engineering experience
• Strong statistical analysis and experimental design capabilities
• Creative problem-solving with robust red-teaming mindset
• Hands-on experience in ML research engineering, stochastic systems, LLM applications, or AI evaluations
• Proven ability to deliver end-to-end projects in fast-paced environments

Preferred:

• First-hand red-teaming experience with complex systems
• Cross-functional collaboration success
• Excellent technical communication skills

Candidates should thrive in ambiguity, demonstrate ownership, and possess the technical depth to tackle frontier AI challenges.

Salary & Benefits

Competitive Compensation: Total compensation for this senior Research Engineer role ranges from $320,000 - $480,000 USD annually, including base salary, equity, and performance bonuses. Exact figures depend on experience and qualifications.

Comprehensive Benefits Package:

• Premium health, dental, vision coverage
• 401(k) with generous company match
• Unlimited PTO policy
• Parental leave and family planning support
• Relocation package for San Francisco
• Weekly catered meals, snacks, and beverages
• Onsite fitness center and wellness programs
• Professional development stipend
• Cutting-edge hardware and AI infrastructure
• Equity in OpenAI - shape the future of AI

Why Join OpenAI?

OpenAI leads the world in developing safe AGI that benefits humanity. Our Frontier Evals team directly influences model training for GPT-4o, o1, GPT-5, and beyond. You'll work alongside brilliant minds, publish influential research, and see your evaluations shape AI's trajectory.

San Francisco location offers unparalleled access to AI talent and innovation ecosystem. Experience rapid model progress firsthand and contribute to humanity's most important technical challenge. OpenAI provides resources, autonomy, and impact unmatched in industry.

Join a mission-driven culture prioritizing safety, rapid iteration, and bold ambition. Your work will define AI evaluation standards for generations.

How to Apply

Ready to push AI safety frontiers? Submit your resume, GitHub/portfolio, and a brief note explaining your fit for this Research Engineer role. Highlight relevant ML research, evaluation experience, or red-teaming projects.

OpenAI is an equal opportunity employer. We celebrate diversity and are committed to creating an inclusive environment for all employees. Applications reviewed on rolling basis - apply immediately to join our frontier team.