Keivalya Pandya
Graduate Teaching Assistant @Khoury College Of Computer Sciences
Signup · Get unlimited contacts
WORK HISTORY
Graduate Teaching Assistant @Khoury College Of Computer Sciences
Boston, MA, US
CS5180/4180 Reinforcement Learning and Sequential Decision Making, taught by Prof Chris Amato and Prof Robert Platt- Develop and maintain RL concept visualization platform (http://cs5180-rl.keivalya.com/) to simulate RL agents in controlled environment- Help students learn the concepts outside the classroom- Encouraged asking questions, focusing on fundamentals, and learning from the first principle method- Support hands-on coding from scratch in exercises, and real-world problem solving by employing RL algorithms.
EDUCATION
Khoury College of Computer Sciences
Master of Science - MS, Robotics
Bright Day School CBSE
Higher Secondary, Science with Mathematics and Computer
Ural State Technical University
Student Exchange, Information Technology
The Coding School
Introduction to Quantum Computing, Quantum Computing
Birla Vishvakarma Mahavidyalaya
Bachelor of Technology - BTech, Mechanical Engineering
ABOUT KEIVALYA PANDYA
I am a robotics and AI researcher working at the intersection of learning, control, and interpretability - building systems that not only act intelligently in the physical world, but whose reasoning we can understand, trust, and improve.My work spans robot learning, reinforcement learning, and vision-language-action models, with a focus on enabling autonomous agents to move from reactive behavior to structured reasoning and robust decision-making. With a foundation in mechanical engineering and a systems-level understanding of robot dynamics and control, I approach intelligence as something that must bridge perception, cognition, and embodiment.I am particularly interested in:> Mechanistic interpretability for embodied AI> Reinforcement learning for long-horizon decision-making> Vision-language-action models for generalizable robot behavior> Bridging model-based control with learned world models> Scaling robot learning from simulation to real-world deploymentMy research philosophy is simple:Intelligence should be grounded, interpretable, and deployable.Across my academic and research journey, I have worked on autonomous manipulation and navigation systems, contributed to reinforcement learning education, and explored how large-scale models can interface with physical agents. I aim to push toward a future where robots do not merely execute policies - they reason about the world, adapt safely, and collaborate seamlessly with humans.I am driven by long-term impact: building embodied intelligence that advances science, industry, and human capability.If you’re working on frontier AI, robot learning, or interpretability for embodied systems - I’d love to connect.
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.