About

I'm an M.Tech student in the Integrated Design & Innovation (iDD) program at IIT Madras. My thesis, advised by Prof. Pulkit Verma, argues that flat-foldable origami tessellations are natively Eulerian, so they give 3D-printer toolpaths their crossing-free, single-stroke structure for free, instead of forcing it with mesh refinement. That work now sits alongside research on coupled robotic dynamics, muscle control, policy distillation, and the evaluation of learned systems.

What ties the projects together

I study how a model’s structure shapes what we can design, control, and learn from it. Sometimes that structure is geometric: a crease pattern or a fabricable linkage. Sometimes it is dynamical: coupling between pneumatic segments, or the switches in a muscle model. In machine learning, I ask what a compact policy preserves and what an evaluation metric can actually establish.

• Origami-arm control combines coupled dynamics with MPC and policy distillation.
• Muscle activation control studies how directional linearizations affect prediction and feedback.
• Interpretability-guided distillation tests response agreement, feature sufficiency, and sampling controls.
• Reasoning evaluation examines what a faithfulness metric measures when transferred to hard mathematics.
• Biohybrid recruitment tests the force-control benefits and physical limits of patterned stimulation.

Build the model, test what it predicts, and let the controls change the conclusion.

Tools I reach for

Elsewhere

I write up longer, more informal versions of this research on the blog. Code for these projects lives on GitHub (repos are being made public on a rolling basis). The fastest way to reach me is email.