Young Investigator, FlexOlmo

Persons in these roles are welcome to work remotely from Berkeley, CA. Compensation: $159,650

Ai2 is seeking talented and motivated Postdoctoral Young Investigator to join the Flex Olmo team
, working on a series of large language models designed for flexible data use
, with a focus on Mixture-of-Experts (MoE),
long-context language models (LCLMs), and retrieval
. Postdoctoral Young Investigators will be based in Berkeley, California
.

This opportunity offers a unique opportunity to contribute to cutting-edge research in natural language processing and machine learning in an exciting, fast-paced research environment. You have the opportunity to:

• Define and lead a high-impact research project.
• Train and release leading models.
• Collaborate with and learn from team members across Ai2.
• Build open-source software for the research community.
• Author scientific papers for publication in a high-profile conference or journal.

• Duration: 1-3 years
• Start date: Flexible
• Candidates: Are within one year of completing their PhD, or already have a PhD
• The Allen AI Young Investigator is a postdoctoral program offering unique benefits. The program will enable you to balance working collaboratively on an Ai2 project while having opportunities to mentor junior researchers.

We design new architectures and training methods that help models use data more effectively—through improved training, inference-time conditioning, and retrieval—broadening the types of data they can leverage and ultimately enhancing performance. We also develop scientific methodologies for evaluating and understanding these systems. Our team produces high-impact research and expertly engineered open-source tools that accelerate NLP research worldwide.

We lead the Flex Olmo project, whose first release in July 2025 focused on a new Mixture-of-Experts architecture. Looking ahead, we plan to pursue creative, groundbreaking research that delivers scientific insights and practical solutions for building architectures and training methods that unlock the use of large and diverse data sources.

Why Flex Olmo? We are building the foundation for research into the next generation of language models designed for flexible data use.

• Flex Olmo is a small, tightly knit team, giving you the unique opportunity to work closely with team members toward one high-impact project.
• We encourage open collaboration projects, even with researchers at external institutions. Team member will be based in Berkeley, with opportunities to engage actively with the University of California, Berkeley, and the BAIR lab.
• Our pay is competitive, and visa sponsorship is available.
• We are committed to open science and support students freely publishing papers, as exemplified by our first release:
  Flex Olmo:
  Open Language Models for Flexible Data Use.

The essential functions include, but are not limited to the following:

• Dedicated Ai2 mentor who is also a faculty member at the University of California, Berkeley.
• 50% work on leading and collaborating on an Ai2 project as an independent contributor (IC).
• 50% work on mentoring junior researchers (PhD students/interns, predoctoral students/interns) as well as opportunities to receive mentorship in academic activities such as grant writing and teaching, if desired.

• Are within one year of completing their PhD, or already have a PhD, in Computer Science or similar field with research experience in machine learning, natural language processing, language and vision, or related areas.
• Outstanding individual contributor (IC) skills
  , especially with deep learning frameworks (e.g. PyTorch).
• An outstanding publication record at AI-related venues, such as NeurIPS, ICLR, ICML, COLM, ACL, EMNLP. We will specifically evaluate the quality of publications in terms of rigor and impact, not the quantity.
• Extensive research experience in areas such as large language models, training dynamics, scaling laws, and data curation. Experience with mixture-of-experts, long-context language models, and retrieval is preferred but not required.
• Located [or willing to relocate]…