Research Scientist • Senior

Who you are

You have a strong background in research, particularly in the fields of machine learning and reinforcement learning — your experience includes developing novel architectures and methodologies that push the boundaries of AI capabilities. You are adept at working with large language models and understand the intricacies of aligning AI systems with human values. Your collaborative spirit shines through as you work closely with cross-functional teams, ensuring that your research translates into practical improvements in production systems.

You possess a deep understanding of reward modeling and have experience in evaluating and grading AI systems — your knowledge extends to identifying and mitigating reward hacking, which is crucial for developing safe and beneficial AI. You are driven by a mission to create interpretable and steerable AI systems that are reliable and beneficial for society. You thrive in environments that challenge you to think critically and creatively about complex problems.

Desirable

Experience with large-scale AI systems and a track record of publishing in reputable conferences or journals would be advantageous. Familiarity with tools and frameworks used in AI research, such as TensorFlow or PyTorch, is also a plus. You are comfortable navigating the challenges of AI alignment and are eager to contribute to the advancement of this field.

What you'll do

As a Senior Research Scientist at Anthropic, you will lead research efforts focused on reward models, shaping how our AI systems understand and optimize for human preferences. Your work will involve developing innovative architectures and training methodologies for reinforcement learning from human feedback (RLHF). You will explore new approaches to evaluation and grading of AI systems, including rubric-based methods, to ensure that our models align closely with human values.

Collaboration is key in this role — you will work alongside teams in Finetuning, Alignment Science, and our broader research organization to ensure that your research leads to tangible improvements in model capabilities and safety. You will have access to cutting-edge models and significant computational resources, allowing you to tackle some of the most pressing challenges in AI alignment.

Your research will not only advance the science of AI but also contribute to the practical deployment of these systems in real-world applications. You will drive ambitious research agendas while also ensuring that your findings are implemented effectively in production systems. This role offers the opportunity to work on critical open problems in AI, making a meaningful impact on the future of technology.

What we offer

At Anthropic, we are committed to creating a supportive and collaborative work environment. We offer competitive compensation and benefits, including optional equity donation matching, generous vacation and parental leave, and flexible working hours. Our office in San Francisco provides a lovely space for collaboration, and we encourage a culture of open communication and teamwork. You will be part of a mission-driven organization focused on building beneficial AI systems that prioritize safety and reliability for users and society as a whole.