Ketan Suhaas Saichandran

WORK HISTORY

EDUCATION

ABOUT KETAN SUHAAS SAICHANDRAN

I\'m an AI researcher and engineer who has worked across the full stack — from publishing in top venues to deploying systems in production- My research covers a lot of ground: multi-agent systems for autonomous scientific discovery, reinforcement learning for active feature acquisition, LLM behavior and cognition, vision-language models for medical imaging, diffusion model alignment, and reasoning evaluation in language models- I\'ve worked with Prof. Mahzarin Banaji (Harvard, pioneering researcher in implicit cognition and one of the most cited social psychologists alive), Dr. Vijaya B. Kolachalama (BU School of Medicine, leading researcher at the intersection of AI and computational medicine), and Dr. Deepti Ghadiyaram (formerly Meta AI, now at Runway & BU)- Published papers in top venues like PNAS, ICML, and CVPR, with several papers under review (ACL, ICML, Nature Partner Journals) and ongoing work targeting Nature Medicine- Our PNAS paper, showing that GPT-4o exhibits humanlike cognitive dissonance, an important observation, was acknowledged by Geoffrey Hinton, the godfather of AI, and was widely covered in media- I gave an oral presentation at CVPR 2025 on improving alignment in text-to-image models- Top 10 Finalist, 2025 Agentic AI Prize for Alzheimer\'s Disease Research- I\'ve also built agentic systems at Clairyon, a UCSD-led clinical AI company, building AI agents for hospital workflows, EHR integration, and clinical decision support- The future of internet is agent-to-agent interaction. No browsers, no websites. Just agents interacting. These agents need verified and consistent identities, like DNS for IP addresses. So I built the United Republic of Agents (https://unitedrepublicofagents.com)- Currently, I\'m building agentic AI systems for enterprise operations at dcortex.