Research Fellow (Physics)

About the Role

The AI for Science Gym is dedicated to fostering AI proficiency throughout the science and engineering community at the National University of Singapore. The Discovery Gym component targets researchers who possess foundational knowledge in data science and machine learning (ML), assisting them in advancing the boundaries of their respective fields.

As a Discovery Gym Lead, you will serve as an AI-empowered data scientist anchored within a scientific area. Your responsibility includes conducting weekly small-group cohorts of postgraduate students and postdoctoral researchers, collaborating with research teams to convert their instrument-derived data into practical AI/ML applications, and developing instructional resources for university-wide dissemination.

Key Responsibilities

• Facilitate weekly peer-learning sessions within your assigned departments, guiding participants through complex problems using their own data and models in an apprenticeship-like format, as well as managing the onboarding of new departments.
• Coordinate one-month Discovery Sprints with domain collaborators, evaluating data set suitability, setting goals and data provenance, producing a functional dashboard within the first week, establishing research objectives and progress checkpoints, concluding sprints by transferring skills and scripts, and curating pedagogical datasets for the Gym gallery.
• Contribute to the communal learning platform by encoding partner data as pedagogical challenges with useful provenance and partial anonymization, and provide demonstrations linking domain-specific workflows with AI methodologies.
• Engage in publishing outcomes and competitive events by leading sprints that produce educational preprints and participating in competitions against other AI meta-frameworks.
• Attend monthly Discovery Gym Lead (DGL) methods-exchange meetings and weekly coordination sessions with the architect and peer leads.

Working Expectations

• Manage 1 to 2 active sprints concurrently along with 1 to 2 additional projects in support; allocate no more than 40% of your workweek to consultation; complete active projects typically within three months; publish at least one scholarly paper or proceeding annually.

Qualifications

• Ph.D. degree in a science or engineering field such as physics, chemistry, biology, materials science, chemical or biomedical engineering, pharmacy, or food science.

Required Skills and Experience

• Comprehensive expertise in data science and machine learning pipelines, including data wrangling, dimensionality reduction, clustering, labeling, and both supervised and unsupervised learning, with an understanding of their limitations.
• Proficiency in scientific Python programming and experience with modern machine learning libraries such as PyTorch or JAX, scikit-learn, pandas, and numpy, combined with reproducible, version-controlled workflows.
• Experience handling complex, real-world instrument data and the ability to assess dataset viability effectively.
• Capability to rapidly prototype functional interactive dashboards with unfamiliar data within short timeframes.
• Strong communication and pedagogical skills, including a critical approach to AI-assisted development and instruction.
• A consistent publication record and motivation to continue scholarly contributions.

Additional Experience Preferences

• Postdoctoral or professional experience applying machine learning in research settings.
• Familiarity with high-performance computing (HPC) environments, multi-GPU setups, and containerized workflows.
• Knowledge of representation learning, foundation or self-supervised models, and physics-informed methodologies.
• Experience designing or evaluating agentic workflows or large language model (LLM) systems.
• Background in mentoring, teaching, or conducting workshops and competitions.
• Cross-domain expertise encompassing multiple scientific disciplines.