Amazon is looking for talented Postdoctoral Scientists to join our team for a one-year, full-time research position. Postdoctoral Scientists will innovate in key areas of machine learning, natural language processing, computer vision, automated reasoning, robotics, or quantum technologies.

AGI Foundations is looking for a highly motivated, ingenious, and versatile Applied Scientist to work with both internal and external science teams to drive innovations that advance the frontiers of artificial intelligence. You will play a central role in the development and execution of new AI challenges that push the limits of large language models and agentic AI systems. A recent example of a challenge is the Amazon Nova AI Challenge: Trusted AI, focusing on responsible AI for LLM-enabled coding assistants.

In this role, you will draw on deep technical experience working with large language models (including pre-training, fine-tuning, and alignment) in order to develop challenge models and tools, mentor and collaborate with participating academic teams on their research, and work closely with internal teams across AGI and Amazon more broadly to apply challenge innovations. Evaluation is also a key aspect of the role. You will design evaluation metrics, conduct experiments, and implement automated evaluation pipelines incorporating LLM-as-a-judge methods.

The ability to communicate in a clear and compelling way is also very important given the high level of both internal and external visibility of your work. You would work with the team to promote new challenges in the external scientific community, present at challenge bootcamps, workshops, and summit meetings, evangelize challenge results and innovations with internal teams, and present at external academic conferences.

A day in the life:
This is a highly varied role with significant exposure and impact both within Amazon and with the external science community. On a given day you might work closely with engineering and product to design and iterate on a new challenge concept. On another you might experiment with different techniques for automated evaluation of LLM outputs or work with the team to plan and train a large language model for use in a challenge. You will also meet with faculty and students to understand their progress and develop collaborations with scientists across the broader AGI org and beyond to apply key innovations.

About the team:
The

Postdoctoral Scientist, AGI Foundations