Role Overview

The Human Data team at OpenAI is at the heart of developing reliable AI systems. We design end-to-end programs that capture the depth of human intent for both everyday and high-stakes AI applications. Our work spans bespoke data campaigns, scalable synthetic data generation, and product-embedded signals. As a Research Engineer, you'll invent novel systems to maximize human feedback utility at the capability frontier.

Based in San Francisco, CA, this hybrid role requires at least 3 days in the office per week. OpenAI provides relocation assistance to help you join our innovative team. You'll partner closely with researchers to translate human signals into training datasets, novel evaluations, and feedback loops that advance our models' capabilities and real-world applications.

This position is ideal for engineers with strong ML backgrounds who thrive on first-principles thinking, experimentation, and shipping impactful systems. Contribute to OpenAI's mission of ensuring general-purpose AI benefits humanity through safe, powerful deployments.

Key Responsibilities

As a Research Engineer on the Human Data team, your work will directly influence model performance and AI safety. Key responsibilities include:

• Partnering with researchers and product managers to design, derisk, and implement groundbreaking 0-to-1 innovations in human feedback capture and processing.
• Uncovering actionable insights from AI trainers and end-users, applying them to core OpenAI research objectives like model scaling and alignment.
• Measuring the efficacy of current feedback methods through rigorous experiments and iterating to boost performance.
• Designing adaptive methods that precisely capture nuanced, subjective human feedback across diverse, context-dependent goals.
• Ensuring evaluations evolve with real-world model usage, accurately tracking utility in products like ChatGPT.
• Running sophisticated online and offline experiments to quantify causal downstream impact on model capabilities.
• Building systems to weight explicit feedback and decode implicit signals from user interactions in OpenAI products.
• Developing scalable pipelines for synthetic data generation augmented by human supervision.
• Collaborating across teams to integrate feedback into RLHF datasets and evaluation benchmarks.
• Optimizing feedback loops for continuous model improvement in production environments.
• Analyzing large-scale trainer data to refine annotation quality and reduce biases.
• Shipping production-ready systems that handle millions of feedback signals efficiently.
• Experimenting with novel supervision techniques to push beyond current SOTA in AI alignment.

Expect to wear multiple hats: from high-level system design to low-level PyTorch implementation, all while maintaining a bias toward measurable impact.

Qualifications

To thrive in this role, you'll need a blend of technical expertise, curiosity, and grit. Ideal candidates:

• Have hands-on experience with ML frameworks like PyTorch and large-scale model experimentation.
• Exhibit first-principles thinking and relentless drive to improve processes, even when challenging.
• Are goal-oriented, embracing tedious high-value work without hesitation.
• Fluidly navigate ambiguous problems, adapting from research ideation to engineering execution.
• Possess exceptional problem-solving skills with a hands-on, action-biased mindset.
• Demonstrate self-reliance, unblocking progress despite external dependencies.
• Are deeply motivated by OpenAI's mission to develop safe AGI benefiting all humanity.
• Bring experience in AI evaluation, human feedback, or data-centric ML research.
• Comfortable with Python, data pipelines, and statistical analysis for experiments.
• Have a track record of shipping ML systems with real-world impact.

Prior exposure to RLHF, synthetic data, or production AI systems is a strong plus.

Salary & Benefits

OpenAI offers competitive compensation for Research Engineers in Human Data roles, estimated at $250,000 - $450,000 USD yearly base, plus equity, bonuses, and comprehensive benefits. Total compensation reflects experience and impact.

Benefits include:

• Top-tier medical, dental, vision insurance.
• 401(k) with generous matching.
• Unlimited PTO and flexible holidays.
• Relocation support to San Francisco.
• Catered meals, wellness stipends, gym reimbursements.
• Parental leave and family planning support.
• Professional development budget for conferences and courses.
• Hybrid SF office with state-of-the-art facilities.

Why Join OpenAI?

OpenAI is the leading AI research lab, creators of GPT models powering tools used by millions. Our Human Data team drives breakthroughs in alignment and capabilities. Work on frontier problems with the best talent, in a mission-driven culture prioritizing safety and humanity.

San Francisco's vibrant tech ecosystem, combined with OpenAI's hybrid model, offers the perfect balance. Contribute to products shaping the future while enjoying Bay Area perks like world-class dining, culture, and innovation hubs.

We're committed to diversity, equity, and inclusion, valuing perspectives from all backgrounds to build AI for everyone.

How to Apply

Ready to invent the future of human-AI feedback? Submit your resume, LinkedIn, and a brief note on why you're excited about this role. Highlight ML projects, especially involving feedback or evaluation.

OpenAI reviews applications on a rolling basis. Top candidates advance to technical interviews covering ML systems, experiments, and problem-solving. Join us in San Francisco to push AI frontiers!