As a third year student, I'm pursuing Electronics and Communications Engineering with a specialization in Data Science. I'm now striving to improve my skills in the field of Computer Vision. I'm currently learning the PyTorch and OpenCV frameworks. I have used PyTorch to implement CNN-based architectures such as AlexNet and LeNet-5, as well as Generative Adversarial Networks. I intend to conduct research on popular variations of Generative Adversarial Networks, as well as diffusion based models such as Denoising Diffusion Probabilistic Models. I am excited to collaborate with those who share my passion and interests.
- B.Tech, Electronics and Communication Engineering with specialization in Data Science | SRM Institute of Science and Technology, KTR (July 2025)
- 12th (ISC Board) | City Montessory School, Kanpur Road, Lucknow (April 2020)
Implemented the famous Generative Adversarial Network architecture proposed in "Generative Adversarial Nets" by Goodfellow et. al in 2014. Frameworks such as PyTorch and tensorboard have been used along with the torchvision library. The model recreates synthetic MNIST data which includes handwritten digits.
Used Colab Notebooks to learn and document my journey of learning PyTorch. It's a collection of interactive notebooks that help in understanding how to use the PyTorch framework with many different use cases such as implementing ANNs, CNNs, Custom Datasets etc.
A simple PyTorch implementation of the paper "Gradient-Based Learning Applied to Document Recognition" by LeCun et. al released in the year 1998. in PyTorch. The model has been implemented on the MNIST handwritten digits dataset and the goal is to classify the image data into respective classes. I have also tried to improve the model using ReLU instead of Tanh and max-pooling instead of average-pooling. The model has an accuracy of 98%.
This is an implementaiton of AlexNet, as introduced in the paper "ImageNet Classification with Deep Convolutional Neural Networks" by Alex Krizhevsky et al. (original paper)
This was the first very successful CNN for image classification that led to breakout of deep learning 'hype', as well as the first successful example of utilizing dropout layers.
This implementation uses the CIFAR10 dataset released by the University of Toronto. The CIFAR-10 dataset consists of 60000 32x32 colour images in 10 classes, with 6000 images per class. There are 50000 training images and 10000 test images. The trained model has an accuracy of upto 76%.
- Associate, McCarthy Lab | NextTechLab, SRMIST, KTR
- Technical Team Member - Machine Learning | Swift Coding Club, SRMIST, KTR