Oct 2018 - Sept 2020
Double-degree masters program in Computer Science specializing in Robotics and Machine Learning,
with a minor in Innovation and Entrepreunership.
Thesis: Bayesian
Optimization for Neural architecture Search
Aug 2014 - May 2018
Bachelor of Technology in Electrical and Electronics Engineering.
Thesis: Self Driving car Steering angle prediction using Android Things
Feb 2021 - Present
I was part of the core development team that introduced a graph based operation framework to work with different ML frameworks like PyTorch and TensorFlow. Our team also handled implenting various adversarial attacks, metrics and also new techniques for synthetic image generation. I have contributed to many external projects in association with various organizations such as Federal Office for Information Security and Fraunhofer Research Institutes.
Feb 2020 - July 2020
Worked on Bayesian optimization for Neural Architecture Search (NAS), where I developed graph kernels to identify optimal architecture using Gaussian modeling. I adapted graph convolutional network as a similarity kernel to identify the best architecture using seen observations. The resulant model was better than current state-of-the-art search techniques on NASBench101 even under hardware constraints. The model achieved over 100x improvement compared to Random Search.
May 2017 - Aug 2017
Developed new state-of-the-art convolutional neural network architectures to accurately segment Optical Coherence Tomograpy(OCT) scans of the Optic Nerve Head(ONH). The network could efficiently segment the 7 important layers of the ONH within a matter of milli seconds. The results obtained are significantly better than any other model in existence (Based on Dice scores). Further improvement in speed can lead to real-time segmentation for use in medical procedures.
May 2016 - Aug 2016
Developed and implemented the parallel algorithm for the granger causality analysis of MRI scans. The parallel implementation of this code was done using Nvidia's CUDA C platform. Achieved significant improvements in execution speed.
Aug 2015 - Aug 2018
Premier computing club of NIT Trichy. Developed many android and web applications for college fest. Organised workshops on programming languages for junior students.
Python
Matlab
Java
C
C++
Bash
CUDA
TensorFlow
PyTorch
NNabla
Keras
CUDA
ROS
Numpy
Scikit Learn
OpenCV