Lead AI Research Scientist, Recursive Self Improvement, AI Safety and Reinforcement Learning

What You Do At AMD Changes Everything

At AMD, our mission is to build great products that accelerate next-generation computing experiences—from AI and data centers, to PCs, gaming and embedded systems. Grounded in a culture of innovation and collaboration, we believe real progress comes from bold ideas, human ingenuity and a shared passion to create something extraordinary. When you join AMD, you'll discover the real differentiator is our culture.

We push the limits of innovation to solve the world's most important challenges—striving for execution excellence, while being direct, humble, collaborative, and inclusive of diverse perspectives. Join us as we shape the future of AI and beyond. Together, we advance your career.

We are hiring a Lead AI Research Scientist, Recursive Self Improvement, AI Safety and Reinforcement Learning focused on recursive self-improvement (RSI) in a bounded, engineering-first sense: systems where models, data generators, or tool chains participate in improving their own training signals, curricula, or verification—always under explicit governance, kill switches, and human oversight. You will research when such loops help (e.g. synthetic data quality, targeted self-play, automated curriculum refinement) versus when they amplify bias or reward hacking, and you will design measurement and containment so RSI-style pipelines remain auditable and safe for AMD's AI-for-HW and generative-AI programs.

You are skeptical by default but constructive: you formalize assumptions, bound autonomy, and insist on counterfactual evaluation. You connect RSI concepts to concrete metrics—data efficiency, robustness, regression rates—not open-ended capability claims.

• Research self-improving training loops: model-generated supervision, iterative distillation, self-critique, and automated curriculum updates with clear scope limits
• Develop theory- and systems-grounded evaluations for capability drift, Goodhart effects, and distributional shift in closed-loop training
• Partner with RL scientists on where RSI-style objectives intersect policy optimization and preference learning
• Define red-team protocols and monitoring for RSI pilots; document rollback criteria before experiments touch shared infrastructure
• Publish or produce technical reports where appropriate; align internal narrative with responsible deployment standards

• Strong background in machine learning (ML), AI safety, reinforcement learning, or a related field, with publications or substantial work in iterative training, self-training, or open-ended learning.
• Experience with empirical safety evaluation, scalable oversight, or stress-testing of generative model training pipelines
• Strong software skills for building controlled experimental harnesses and reproducible RSI microcosms

• PhD in Computer Science, Machine Learning, or related field strongly preferred.