rohit khadka

I'm Rohit, a Computer Science student who enjoys thinking about how intelligence actually works—in machines, in markets, and occasionally in humans. I build AI systems not just to make them perform well, but to understand why they behave the way they do (and why they sometimes behave badly).

My work lives at the crossroads of AI, mathematics, and finance. I've built machine learning and deep learning projects using Python, PyTorch, and scikit-learn, explored trading strategies through backtesting, and worked on research projects like a multi-LLM framework for creative mathematics problem generation. I like turning abstract ideas into code—and then stress-testing my own assumptions until something breaks, preferably in simulation.

I enjoy both theory and practice. Mathematics feels like the cleanest language we have for describing reality, while AI feels like a slightly opinionated assistant that learns from examples and occasionally overfits to life. I care about clarity, rigor, and building models that behave sensibly outside ideal conditions—because the real world rarely reads the documentation.

At the moment, my research focus is on exploring how macroeconomic shocks and financial signals can be modeled using a combination of econometric methods (e.g., VAR frameworks) and machine learning approaches (e.g., LLM-based information extraction and agent-based simulations). I'm drawn to problems that are a bit messy, a bit uncomfortable, and usually more interesting than they first appear.