AI Research Engineer, Representation Learning

What you'll do

• Develop and experiment with transformer-based models for structured and credit time-series data.
• Analyze learned representations to uncover how the model encodes complex financial concepts.
• Conduct rigorous experiments to validate hypotheses on model behavior and training dynamics.
• Train, evaluate, and debug deep learning models using PyTorch/TensorFlow to create prototypes.
• Extend core models into broader discriminative and generative modeling architectures.
• Maintain expertise in modern deep learning techniques including sequence-to-sequence and diffusion models.

What we're looking for

• PhD in ML/AI/CS/EE or related field with 3+ years of experience, or MS with 5+ years.
• Strong foundation in Transformer architectures, attention mechanisms, and sequence modeling.
• Deep knowledge of linear algebra, statistics, probability for characterizing model behavior.
• Experience analyzing learned representations (latent spaces, embeddings) from deep learning models.
• Technical intuition for deep learning training dynamics including stability and gradient behavior.
• Ability to write clean, efficient Python code for developing research prototypes.