Project Overview

We are seeking a research intern to explore the development of AI-based tutoring technologies to support interactive learning. One such technology involves AI systems that assist in interactive question answering and step-by-step explanations for Mathematics for primary and secondary school students. Research will involve leveraging and customising existing foundation models for problems that require both image and text understanding such as Geometry. In particular, we are interested in combining these models with classical tools such as solvers and verifiers, to ensure greater rigour and accuracy in the generated problem, solutions and explanations.

Responsibilities

• Incorporate tools such as solvers or verifiers to generate new maths problems or improve the reasoning capabilities of LLMs
• Extend the experimental pipeline to problems with visual context e.g., plane geometry.
• Understand the failure modes of existing LLMs on maths problems and identify areas where neuro-symbolic methods can boost performance.
• Incorporate these tools into intelligent tutoring systems that can identify leaner’s knowledge gap and generate the most relevant questions or hints.

Requirements

• Currently pursuing a Bachelor or Master degree in Computer Science, Mathematics, Machine Learning or other related fields.
• Proficiency in programming languages like Python and deep learning libraries like PyTorch.
• Experience working with (e.g., training, adapting, evaluating) foundation models.

Preferred Qualifications

• Experience with using SMT solvers, formal verifiers or other math tools.
• Familiarity with personalized education approaches.
• Contribution to publications at top-tier ML conferences or journals or to open-source projects.

This internship is only open to full-time matriculated or registered students in the six autonomous universities Singapore. The internship must be approved by the University.