Mohan Kumar Srirama

Mohan Kumar Srirama



Hello there!

I’m an AI Researcher at Skild AI. I work on building foundation models for robots that let them generalize easily to new scenes and operate robustly in the wild.

I recently graudated with a Robotics Master’s at Carnegie Mellon’s School of Computer Science, where I was fortunate to be advised by Professor Deepak Pathak and Professor Abhinav Gupta . My research focused on developing visual representations that blend insights from both human and robot data to accelerate robot skill learning.

Previously, I was a Senior Camera Systems Engineer at Qualcomm R&D building algorithms for Image Signal Processors (ISPs) in Snapdragon🐉 chipsets that power over a billion smartphones worldwide.

Beyond research, I enjoy lifting weights 🏋️‍♂️, mastering dynamic poses 🤸, and catching the latest Formula 1 race 🏎️. I’m always open to collaborating or just geeking out over cool ideas!

Email: mohankus [at] cs [dot] cmu [dot] edu

कर्मण्येवाधिकारस्ते मा फलेषु कदाचन।
मा कर्मफलहेतुर्भूर्मा ते सङ्गोऽस्त्वकर्मणि॥

— Bhagavad Gita, Chapter 2, Verse 47

News


Publications

Policy Agnostic RL Fine-Tuning Multiple Policy Classes with Actor-Critic RL
ICLR Generative Models for Robot Learning Worshop, 2025   [Best Paper Award]
Policy Agnostic RL Fine-Tuning Multiple Policy Classes with Actor-Critic RL
Bimanual Dexterity for Complex Tasks
Conference on Robot Learning (CoRL) 2024
Bimanual Dexterity for Complex Tasks
DROID: A Large-Scale In-The-Wild Robot Manipulation Dataset
Robotics: Science and Systems (RSS), 2024  [Oral Presentation]
DROID: A Large-Scale In-The-Wild Robot Manipulation Dataset
Train Offline, Test Online: A Real Robot Learning Benchmark (TOTO)
IEEE International Conference on Robotics and Automation (ICRA), 2023 [Best Paper Award at NeurIPS WBRC 2022]
Train Offline, Test Online: A Real Robot Learning Benchmark (TOTO)

Projects

*
manimo

manimo

A modular interface for robotic manipulation.

tire-defect-detector

tire-defect-detector

An Efficient Net based CV pipeline for tire defect detection delivered to a top global tire manufacturer.

Recent & Upcoming Talks