Kireeti
I am a recent graduate in Computer Science and Engineering with an interest in Python, Rust, and machine learning. A few years ago, I developed an interest in machine learning, and since then I have been self-learning, building projects, and writing about what I learn. During my undergraduate studies, I worked across different areas of machine learning, primarily on federated learning under the supervision of Dr. Karthick S. I have broad interests across ML, and my current focus is on distillation techniques and federated learning.
Beyond ML research, I also enjoy software development work, where I have implemented several systems during my previous internships. I started this blog as a way to share my thoughts and opinions. Outside programming, my interests include manual cars, music, anime, and literature(I am trying to get back into reading more).
Projects
Some of my older projects are not structured as cleanly as I would like yet, so they may be difficult to follow. I am keeping them here as a map of what I have explored and as placeholders for future writeups.
Federated Learning
FL was the topic of my bachelor’s thesis and took up most of the second half of my undergraduate studies. I implemented and compared several FL algorithms, including FedAvg, FedProx, and IFCA, and benchmarked their performance under different data heterogeneity scenarios. I also had the opportunity to work under Dr. Karthick S on developing a novel FL algorithm, which has been published in the journal Neural Computing and Applications. Refer to this post for an introduction to FL.
RTDETR
RTDETR is an object detection model that uses components such as a hybrid encoder and AIFI to reduce inference time. This was part of the work I previously did at a company, where RTDETR was used alongside YOLO-seg and other image processing modules for an end-to-end computer vision pipeline. This repository implements the RTDETR model and provides scripts to train and distill it using standard distillation techniques [1], [2].
Other Projects
- ChessAI
- Graph Neural Networks
- Open-set Face verification
- PINNs for Harmonic Oscillator
- Image colorization using GANs
- A* path finding visualizer
Thanks for reading!