Pre-Doctoral Fellow: Reinforcement Learning for Large Scale Foundation Model Post-Training

School: Harvard John A. Paulson School of Engineering and Applied Sciences

Department/Area: Computer Science

Position Description

We invite applications for a Fellow to work with Assistant Professor Kiante Brantley at Harvard SEAS. The lab's research focuses on improving the capabilities of foundation models using reinforcement learning-from theory to practice. This position will emphasize large-scale reinforcement learning for post-training, studying both practical and theoretical issues pertaining to post-training techniques.

Successful candidates must be able to work as part of a team performing research to support the project. Additional responsibilities include preparing publications and reports, mentoring graduate students, and collaborating with other groups. Successful candidates will be self-motivated, have a strong work ethic, be technically skilled, and have strong oral and written skills. Start dates are flexible, but the positions will be filled as soon as possible and will initially be for 6 months, renewable for an additional 6 months following performance review and funding availability.

Basic Qualifications

BA/BS in computer science, applied physics, applied math, or a related discipline.

Additional Qualifications

Individuals with a demonstrated track record in scientific research, which can be evidenced through publications, technical reports, or impactful software projects.

Special Instructions

A complete application must include a curriculum vitae, 2-5 letters of reference, 2 publication samples, and a 1-2-page research statement. In your cover letter, please also address the following supplemental questions: (1) What sorts of projects would you like to carry out in the lab? (2) What are your greatest strengths, and what areas would you like to improve?

Contact Information

Anana Charles, Faculty Coordinator

Contact Email: acharles@seas.harvard.edu

Salary Range

$50,000 - $70,000

Pay offered to the selected candidate is dependent on factors such as rank, years of experience, training or qualification, field of scholarship, and accomplishments in the field.

Minimum Number of References Required: 2

Maximum Number of References Allowed: 3

Keywords

Machine Learning
Reinforcement Learning

Foundational Models
Post-Training
Large Language Models