Postdoctoral Research Scientist
Job Summary
A postdoctoral research scientist position is available in the Department of Biomedical Informatics (DBMI) at the Columbia University Irving Medical Center. We are looking for a postdoc research scientist with a background in computer science and machine learning, and expertise in reinforcement learning.
The postdoc will be joining a project funded by the National Science Foundation titled “Human-Centered Reinforcement Learning for Personalized Nutritional Coaching”. The project will develop a novel approach to human-aligned reinforcement learning to learn dialog structures for health coaching that balance optimal collection of information with intuitiveness and logical conversational structures as perceived by humans.
Duties and Responsibilities
The post-doctoral research scientist will work in collaboration with the research team to investigate human-centered and scalable reinforcement learning (RL) solutions to learn conversational structures that align with human expectations. This will include working with a multidisciplinary team to develop a corpus of dialogs to identify human preferences and expectations for optimal dialog structures, developing novel RL architectures, and working with a team to develop a chatbot that incorporates this novel RL architecture.
The research will cover methodological and applied aspects of reinforcement learning, inverse reinforcement learning, and aligning RL with human preferences, natural language processing and human-computer interaction.
The postdoctoral research scientist will be provided with methodological and applied mentoring, and with plenty of opportunities to advance their research experience. This will include preparation of scientific manuscripts and presenting them at conferences and other scientific venues, participating in grant proposal preparation, teaching and mentoring PhD and master’s students, and preparation of new courses, among others. The postdoc will be co-mentored by the co-Principal Investigators, Drs. Urteaga and Mamykina, who will meet weekly to discuss goals and progress.
Qualification
Requires a PhD in Computer Science or related fields.
Preferred Qualifications
- Experience with inverse reinforcement learning (IRL) and learning from human preferences through reinforcement learning (RLHF).
- Background with natural language processing techniques.
- Previous experience with applications of ML in healthcare.
- Previous experience in Human-computer interaction.
Additional Requirements
- Must have excellent organization, oral communication, and scientific writing skills.
- Background and expertise in machine learning and reinforcement learning.
- Experience implementing ML&RL algorithms in Python.
Interested and qualified candidates should submit a CV, cover letter, statement describing their research interests and three letters of references to Columbia University Academic careers website. Quick link: https://academic.careers.columbia.edu/#!/130484
Requisition number 130484.
