Postdoctoral Associate AI Performance and Safety
MIT FutureTech is an MIT interdisciplinary research group studying technical and economic progress in computing and AI.
About MIT FutureTech
An MIT-based research group that produces research on computing and AI progress, scientific and economic impacts, and AI-risk management. It operates public AI-risk data and navigation tools and conducts research, convening, and policy-facing work.
Skills
Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will design, implement, and analyze machine-learning experiments on AI performance, efficiency, safety, robustness, and generalization. You will develop agent evaluation arenas and alignment-evaluation tools, contribute to research and policy writing, and participate in workshops and external collaborations.
Requirements
- PhD in Computer Science, Machine Learning, AI, or a related field
- Deep learning
- TensorFlow
- PyTorch
- Software engineering
- Model training
- Fine-tuning
- Reinforcement learning
- Retrieval-augmented generation
- AI safety
- AI alignment
- Interpretability
- Multi-agent system
- Empirical research
Responsibilities
- Design and conduct AI experiments on performance, robustness, safety, and generalization
- Develop agent evaluation arenas
- Build tools and frameworks for evaluating AI alignment techniques
- Integrate AI safety insights into research and policy recommendations
- Contribute to research papers, white papers, and public reports
- Participate in workshops, conferences, and external collaborations
Hiring Process
Selected candidates will first be interviewed via Zoom.