Teaching networks the physics of money
Most of my work sits where rigorous mathematics meets a messy real world - turning equations into models that actually run, and results into something a non-specialist can act on.
Research Intern, IISc Bangalore
Indian Institute of Science · Bangalore, India
I build physics-informed neural networks that solve the partial differential equations behind option pricing - extending the classic framework to Merton’s jump-diffusion model, where prices can leap rather than drift smoothly. The aim is a method that respects the underlying mathematics while staying fast and flexible enough to be useful in practice.
The three moving parts
Physics-Informed Neural Networks
Neural networks that bake the governing differential equation directly into the loss function, so the model learns the physics of the problem instead of just memorizing data points.
Merton's Jump-Diffusion
An extension of Black-Scholes that adds sudden jumps to price paths on top of smooth diffusion - far closer to how real markets behave around news and shocks.
Option Pricing
Putting both together to value derivatives in regimes where the smooth, continuous assumptions of classical models quietly break down.
MSc, Data Science - VIT Vellore
A degree built on statistics, machine learning, and applied mathematics - the foundation the research stands on. Coursework that rewards getting the fundamentals exactly right before reaching for the clever method. Next stop: an analyst role at Solytics Partners in Pune.