Sarmad Mehrdad
Robotics researcher in inverse reinforcement learning, learning from demonstration, and embodied intelligence.
Hi. I am Sarmad Mehrdad.
I recently completed my Ph.D. in Electrical Engineering at New York University. My work sits at the intersection of robot learning, optimal control, and human motion understanding. I care most about learning methods that are data-efficient, interpretable, and grounded in the behavior of real embodied systems.
What I work on
My research revolves around three broad threads: learning human intent, learning from demonstration, and learning about humans from data.
What I care about
I am drawn to problems that are mathematically grounded, relevant to real robots, and interpretable enough to explain why a behavior emerges.
Where I am headed
Long term, I want to build an academic career in robotics and contribute to a style of robot learning that works with people in a thoughtful and human-centered way.
In a nutshell
My research has centered on three broad threads:
- Learning human intent through inverse reinforcement learning and cost-function inference.
- Learning from humans through demonstration-driven robot learning.
- Learning about humans by modeling human state, motion, and behavior from data.
What I am trying to build toward
Going forward, I want to keep building in the broad space of robot learning for human-centered robotics. What excites me most is the point where learning, control, and embodiment meet.
I like research that does not only optimize performance, but also helps us understand what the model is doing, what assumptions it is making, and what part of the behavior is genuinely meaningful.
A more human note
I want this website to feel like it was written by a real person. So while this page is here to present my work, it is also here to show a bit of the person behind it: what I value, what kind of questions pull me in, and the direction I hope my research life grows into.
If you would like the more formal version, the CV and Research Statement tabs are better fits.
If you would like the more personal version, the Personal tab is where the site loosens up a little.