
About meBio
Robots can see the world, but they can't yet feel it. I work on dexterous manipulation with a focus on touch, which I believe is key to robots that generalise and, one day, use their hands in the real world as naturally as humans do.
I'm a Research Assistant in the SLMC Group at the University of Edinburgh, working with Prof. Sethu Vijayakumar, Victor Leve and Kawada Robotics on getting their NEXTAGE robot to fold cardboard boxes that behave differently from batch to batch, by estimating each box's physical properties from force/torque, vision and touch.
Before this, I did my MSc in AI at Edinburgh with Elle Miller and Sethu, where I trained a Shadow Hand Lite to spin two Baoding balls blind and transferred the policy zero-shot from simulation to the real hand (No Eyes, No Problem; roto 2.0). I did my BTech in IT at NITK Surathkal, where my final-year project with Prof. Anand Kumar M became a COMSNETS paper. I also did research at IIT Roorkee's CANDLE lab with Prof. Sparsh Mittal, which led to a single-author GCPR paper, and at NITK with Prof. Jeny Rajan.
Outside the lab: long walks, one coffee too many, and Suits on its hundredth rewatch.
NewsUpdates
- Oct 2026 Started as a Research Assistant in the SLMC Group, working with Kawada Robotics on box folding with the NEXTAGE robot.
- Aug 2026 Submitted my MSc dissertation, Manipulation in the Dark, and completed my MSc in AI at the University of Edinburgh.
- 2026 roto 2.0 — The Robot Tactile Olympiad accepted at the ViTac workshop, ICRA 2026.
- Jan 2026 Started my MSc dissertation on blind sim-to-real Baoding with the Shadow Hand Lite.
- Jan 2026 Paper published at COMSNETS 2026 on multimodal RL for game recommendation.
- Sep 2025 Started my MSc in Artificial Intelligence at the University of Edinburgh.
- 2025 Paper published at DAGM GCPR 2025 on Mpox lesion classification.
- 2025 Graduated from NITK Surathkal (BTech, Information Technology).
PublicationsResearch
A Cascaded Dilated Convolution Approach for Mpox Lesion Classification
Multimodal Reinforcement Learning for Game Recommendation using Dual-Discriminator PPO Framework
Side Quests & ThesesWork

I Never Showed It the Task
Phone clips of single skills teach a simulated Franka Panda combined tasks it never saw demonstrated.
Side project
Tendon Slack & Actuator Nets
Finding tendon slack in the Shadow Hand Lite and trying three ways to learn it from hardware data.
MSc research
Tabular RL on a Robot I Built
Ten Sutton & Barto algorithms on a custom robot in PyBullet.
Side project
Manipulation in the Dark
MSc dissertation: blind sim-to-real transfer of a Baoding policy on the Shadow Hand Lite, with 55 rotations in a row on the real hand.

Spot the Pox
A cascaded attention module for telling Mpox apart from look-alike rashes, built at IIT Roorkee and later extended solo into a GCPR paper.
Research internship · IIT Roorkee
RL-Powered Generative Recommendations
B.Tech major project with Siddharth Kelkar: sequential game recommendation with PPO and discriminators. It became the COMSNETS paper.

Finding the 0.01%
Tumour segmentation in PET/CT scans where under 0.01% of pixels are tumour, using a custom loss that took Dice from 0.41 to 0.656.
Research project · NITK