Interpretability Researcher

About the role

Interpretability Researchers develop methods to understand and explain AI decisions. This position focuses on investigating how neural networks process information and make predictions, contributing to the scientific community through open‑source tools and publications.

Key responsibilities

• Plan and run experiments: design studies to identify how models process information and make predictions.
• Develop tools: create and maintain open‑source interpretability libraries and visualizations.
• Review literature: stay up‑to‑date on academic research and synthesize findings into internal reports.
• Perform mathematical research: investigate inductive biases and underlying principles of neural networks.
• Communicate findings: write papers, present results, and coordinate with volunteers or collaborators.

Required profile

• Doctoral degree in machine learning, applied mathematics, or a related field.
• Experience in interpretability research and familiarity with tools such as Integrated Gradients, SHAP, or attention visualizations.

Required skills

• Integrated Gradients
• SHAP
• Attention visualizations