Postdoctoral AI Fellow, Post-Training for Materials Discovery

Job description

Job responsibilities

• Define and pursue an original research direction in model post-training, using materials discovery as a scientifically meaningful testbed.
• Develop and evaluate new post-training methods, which may involve reinforcement learning, data generation and selection, evaluation, or agentic workflows.
• Build robust, reusable research prototypes that extend beyond proof of concept.
• Publish research using open-data use cases and present findings at major AI and scientific conferences.
• Apply capabilities to Toyota-relevant use cases, while respecting confidentiality and publication constraints.
• Contribute expertise to related efforts across AMDD.

Job requirements

• Ph.D. in machine learning, computer science, statistics, or a related field, obtained within the past three years.
• Expertise in model post-training or a closely related area, such as reinforcement learning, supervised fine-tuning, evaluation, or agentic systems.
• Strong record of original AI or machine learning research, demonstrated through publications, open-source work, or other significant research contributions.
• Ability to identify important research questions, turn open-ended ideas into well-scoped projects, and design rigorous experiments and evaluations.
• Demonstrated ability to design, implement, and debug readable, modular, and reproducible machine learning research code in Python, with experience in PyTorch or a comparable framework.
• Thrives in a culture that values diversity, collaboration, humility, and learning.
• Experience applying machine learning to scientific or engineering problems or collaborating with domain experts.