Research Fellow in Deep Reinforcement Learning

RA in Deep Reinforcement Learning

The candidate will conduct research work on deep reinforcement learning and AI alignment in the context of large models. One particular focus will be reinforcement learning from human feedback. The ultimate goal is to design a novel technique that can enhance AI alignment. These achievements are expected to be published in top AI or machine learning conferences or journals.

Reports to

Prof. Paul Weng, Associate Professor of Electrical and Computer Engineering 

Essential Duties

The expected tasks to be conducted by the candidate include:

• Research activities.
  • Literature review on related work (i.e., read, summarize, present recent papers)
  • Reproduce experimental results of some previously published papers
  • Implement novel algorithmic ideas
  • Run experiments and analyze experimental results
  • Write reports under the form of a research paper
• Supervision of undergraduate and graduate students.
• Funding applications.
• Other tasks as assigned.
• Work toward achieving assigned annual objectives and KPIs, demonstrating accountability for performance and results.
• University employees' job responsibilities will continue to expand in scope and depth as the University grows in size and complexity in its programs.

Required Qualifications

• The candidate should hold a bachelor or master’s in computer science, AI, machine learning, or mathematics.
• The candidate is expected to be familiar with most of the following topics (or willing to self-learn a novel topic quickly):
• Python programming
• Basics of machine learning and deep learning
• Pytorch or TensorFlow
• (Deep) reinforcement learning
• Pre-trained large models
• AI alignment
• The candidate should be self-motivated, hard-working, rigorous, open-minded, and have an excellent English (written and oral).

Education

The candidate should hold a bachelor or master in computer science, AI, machine learning, or mathematics.

Years of Related Working Experience

Excellent fresh graduate or has 1-2 years relevant research experiences are considered.