Researcher, Alignment

Details

About the Team

The Alignment team at OpenAI is dedicated to ensuring that AI systems are safe, trustworthy, and consistently aligned with human values, even as they scale in complexity and capability. The team develops methodologies that enable AI to robustly follow human intent across adversarial and high-stakes scenarios.

The team's approach has two pillars:

1. Harnessing improved capabilities into alignment, ensuring that alignment techniques improve rather than break as capabilities grow.
2. Centering humans by developing mechanisms and interfaces that enable people to express their intent and supervise and control AI systems, including in highly complex situations.

About the Role

As a Research Engineer or Research Scientist on the Alignment team, you will help ensure that AI systems consistently follow human intent in complex and unpredictable scenarios. You will design and implement scalable solutions for AI alignment and integrate human oversight into AI decision-making.

The role is suited to someone who can move from an ambiguous model-behavior question to a concrete experimental setup: formulate a hypothesis, build an evaluation or intervention, run the experiment, analyze the result, and determine what the evidence supports. This role may be based in San Francisco or London, subject to team needs and location approval.

Responsibilities

• Develop and evaluate alignment capabilities that are subjective, context-dependent, and difficult to measure.
• Design evaluations to reliably measure risks and alignment with human intent and values.
• Build tools and evaluations to study and test model robustness in different situations.
• Design experiments to understand how alignment scales as a function of compute, data, context and action lengths, and adversarial resources.
• Design and evaluate new human-AI interaction paradigms and scalable oversight methods that redefine how humans interact with, understand, and supervise models.
• Train models to be calibrated on correctness and risk.
• Design novel approaches for using AI in alignment research.

Requirements

• A PhD or equivalent research experience in computer science, computational science, data science, cognitive science, or a similar field.
• Strong engineering skills, particularly in designing and optimizing large-scale machine learning systems, such as PyTorch.
• Deep understanding of the science behind alignment algorithms and techniques.
• Ability to develop data visualization or data collection interfaces using technologies such as TypeScript and Python.
• Willingness to perform a variety of tasks that help move the team forward.
• Interest in fast-paced, collaborative, cutting-edge research environments.
• Commitment to developing AI models that are trustworthy, safe, and reliable, especially in high-stakes scenarios.

Benefits

• Equity, performance-related bonuses for eligible employees, and benefits including medical, dental, and vision insurance.
• Pre-tax accounts, a 401(k) retirement plan with employer match, paid parental and caregiver leave, paid time off, company holidays, and paid sick or safe time.
• Mental health and wellness support, employer-paid basic life and disability coverage, and an annual learning and development stipend.
• Daily office meals and meal delivery credits where eligible.
• Relocation support for eligible employees.
• Additional taxable fringe benefits may be provided, including charitable donation matching and wellness stipends.

OpenAI is an equal opportunity employer and provides reasonable accommodations to applicants with disabilities. Background checks are administered in accordance with applicable law.