Nachiket Kotalwar
PhD student · Carnegie Mellon University
I am a PhD student at the Language Technologies Institute at Carnegie Mellon University, advised by Prof. Carolyn Rosé.
My research focuses on human-AI interaction and the capabilities and reliability of AI agents. My goal is to enable effective human-AI collaboration.
Previously, I was a pre-doctoral researcher at Max Planck Institute for Software Systems, advised by Prof. Adish Singla.
If you’re getting started in research or thinking about graduate school, feel free to email me. I’m happy to share my experiences and help you think through next steps.
Research
All publications →Human-AI interaction.
I study how AI systems can adapt to people and provide useful assistance as they work. My research examines how to personalize responses with limited user input, how access to a person’s ongoing work informs when and how AI should help, and how people and agents coordinate in shared tasks. I build and evaluate interactive systems to investigate these questions.
- Visual Access & Proactive Tutoring (preprint, 2026)
- Inference-Time Personalized Alignment (NeurIPS 2025)
- Hints-In-Browser (NeurIPS 2024)
Agent capabilities and reliability.
I study how to build agents that solve complex tasks reliably. My work develops training methods for transferring skills to new tasks and agent systems that combine structured search, tool use, and verification through execution.
- Hybrid-Gym (ICML 2026)
- Ulysses (preprint, 2026)
- ABSENTIA (preprint, 2026)
Applications of my work include AI-assisted software engineering and STEM education.