Postdoctoral Fellow Multimodal Machine Learning

Develop multimodal machine learning methods at McGill University in Montreal, integrating cognitive, conversational, and biometric data through self-supervised learning.

As a Postdoctoral Fellow, you will build a foundation model of human neurotypes from diverse individual-difference data. Your research will combine tokenization, multimodal architectures, latent-space modeling, and psychometric analysis. The work includes evaluating robustness, missing modalities, participant-level separation, and generalization while examining bias and sensitive inferences.

Key Responsibilities:

• Develop tokenization and self-supervised methods for diverse data streams
• Train multimodal architectures mapping data into unified latent spaces
• Distinguish cognitive variation from measurement and language confounds
• Design evaluations for robustness, separation, and generalization
• Scrutinize bias, sensitive inferences, stability, and responsible data usage

Requirements:

• PhD in machine learning, computer science, or related discipline
• Proficiency in Python and modern deep learning frameworks
• Experience with high-performance computing infrastructure and GPU clusters
• Research background in multimodal modeling and representation learning
• Experience with statistical machine learning model evaluation

Apply multimodal modeling, representation learning, and rigorous evaluation to neurotype research at McGill University in Montreal.