Before I joined IIIT Hyderabad, I was a Research Fellow at iHub Data, where I worked with Prof. Bapi Raju S on developing a low-parameter model, LRH-Net, designed to detect multiple fatal cardiovascular diseases at once in resource-constrained environments. Also, through proposed multi-level knowledge distillation, I was able to reduce the required input leads to two or three (generally twelve), enhancing its user-friendliness and suitability for edge devices.
I completed my Bachelors from GGSIPU University Delhi in Information Technology. I worked in the areas of evolutionary algorithms, detection and prediction of COVID-19 using deep learning, and interned at The IIT-BHU - Indian Institute of Technology (BHU) Varanasi with Prof. Hari Prabhat Gupta on quality assessment of River Ganga using machine learning.
LRH-Net: A Multi-level Knowledge Distillation Approach for Low-Resource Heart Network
, Swathi Guptha, Likith Reddy & Bapi Raju
MICCAI Workshop: FAIR'22
Paper /
Code
International Institute of Information Technology
Research Fellow, Healthcare & Artificial Intelligence (HAI)
Jan 2020 - May 2021
International Institute of Information Technology Jan 2020 - May 2021
Research Fellow, Healthcare & Artificial Intelligence (HAI)International Institute of Information Technology Jan 2020 - May 2021
Research Fellow, Healthcare & Artificial Intelligence (HAI)
International Institute of Information Technology, Hyderabad
Master of Science by Research in CSE | CGPA: 9.2 | July'22 – Present
Manipal Institute of Technology, Manipal
Bachelor of Technology in CSE | July'17 – July'21
AutoSub
- Developed a CLI application to generate subtitles for video files on-device automatically
- Implemented MFCC features to segment audio on non-speech segments and perform speech recognition
- Improved performance using an external scorer (language model) and added support for GPU-based inference
from scratch
minimal implementations from scratch of the following:
- baby86: a minimal x86 "bootloader" to print stuff on screen
- nn: dense neural network with multiple layers and activation functions
- cnn: NumPy-only CNN with Conv and MaxPool layers
- torch: not-so-minimal implementation of the torch API
Bioactivity Prediction
- Used regression models to predict biological activity (pIC50 values) of protein targets from ChEMBL database
- Calculated Lipinski and PaDEL descriptors using Acetylcholinesterase (AChE) as the target protein
- The best Decision Tree Regressor model achieved an R-squared value of 0.86
Augmented Random Search for Data Augmentation
- Improved AutoAugment by replacing the discrete search space with continuous space for augmentation policies
- Used Augmented Random Search method to improve performance and maintain diversities between sub-policies
Antenatal Care (iOS App)
- Created an iOS application using Swift and XCode to provide antenatal care for rural populations
- Implemented NFC to store electronic health records like test results, scans, prescription details on an NFC-enabled card
- Used Firebase as a back-end database for storage and retrieval of patient details