Hi there, I'm Rishub Jain!

I was a Research Engineer at DeepMind in the London office, working on Responsible AI. I'll be starting an AI Safety non-profit in August around Amplified Oversight.

Current work:

• Backstory: Improving the Information Quality of the Internet
• Human-AI Complementarity for Identifying Harm

Some past work:

• Human-AI Complementarity: A Goal for Amplified Oversight
• AlphaFold
• Feature Flow for Frame Interpolation
• Improving Single-GPU Performance for DQNs
• Skill Trees for Hierarchal Reinforcement Learning
• RoboCup SSL 2018
• Autonomous Object Translation from Language
• RL-based AI for Breakout and Tetris
• Terminal Competition Project Report

Papers:

• Toward Human-AI Complementarity Across Diverse Tasks (Arxiv 2026)
• Human-AI Complementarity: A Goal for Amplified Oversight (FAccT 2026)
• Can AI Mediation Improve Democratic Deliberation? (Knight First Amendment Institute 2025)
• An Approach to Technical AGI Safety and Security (arXiv 2025)
• Accurate structure prediction of biomolecular interactions with AlphaFold 3 (Nature 2024)
• Gemini: A Family of Highly Capable Multimodal Models (arXiv 2023)
• Highly accurate protein structure prediction for the human proteome (Nature 2021)
• Protein complex prediction with AlphaFold-Multimer (bioRxiv 2021)
• Highly accurate protein structure prediction with AlphaFold (Nature 2021, Nobel in Chemistry 2024)
• Learning and Reusing Dialog for Repeated Interactions with a Situated Social Agent (IVA 2017)
• Creating Prosodic Synchrony for a Robot Co-player in a Speech-controlled Game for Children (HRI 2017 Best Technical Paper)