2nd year CSE, JIIT Noida

Vaibhav Patidar

I write backpropagation by hand, run Monte Carlo simulations in C++, and build clinical risk models. If a library does it, I want to understand how it works before I use it.

96.57%
Test accuracy, MNIST. NumPy neural network, no ML libraries.
87.71%
Accuracy, Brain MRI tumor classification. ROC-AUC 0.9711.
97.38%
Accuracy, Sepsis early warning system. ROC-AUC 0.9238.
0.1%
k-eff convergence, Monte Carlo neutron transport. 10,000 simulations.
GSSoC '26
Open source contributor.
Vaibhav Patidar

About Me

I am a second year Computer Science student at Jaypee Institute of Information Technology, Noida. I prefer building systems from first principles over importing libraries. When I implemented a neural network in pure NumPy, I wrote every forward pass, every gradient, every weight update by hand. That is how I learn.

My current work sits at the intersection of deep learning and healthcare. I have built a brain MRI tumor classifier using transfer learning on ResNet-50, and a sepsis early warning system that predicts onset from ICU time-series data on a 43:1 imbalanced dataset.

Longer term, I am interested in quantitative finance and stochastic modeling. My Monte Carlo neutron transport simulation in C++ reflects that direction: probability theory, numerical methods, and large-scale stochastic computation.

Right now I am focused on PyTorch, reinforcement learning, competitive programming in C++, and deepening my understanding of probability and linear algebra.

/* skills */

Languages
Python C++ C JavaScript HTML/CSS SQL
ML / AI
PyTorch NumPy Pandas Scikit-learn Matplotlib OpenCV Gradio
Concepts
Deep Learning Transfer Learning Neural Networks Backpropagation Monte Carlo Methods Feature Engineering Data Structures Algorithms OOP
Tools
Git / GitHub VS Code Google Colab Linux