Badges
Certifications
Work Experience
Graduate Research Assistant
University of Southern California•  June 2022 - Present
Pioneered a creative Pytorch network architecture for power-map estimation, achieved 300% better results than previous state-of-the-art models. Working on CNN, UNets, Deep lab and Deep Prior by exploiting Google Colab and using Matplotlib and Tensor-board to draw convergence plots for analysis. • Devised a pipeline to convert raw machine text data of over 9 million samples into readable CSV, later converted into images utilizing PyCharm. Potentially would save more than $500 Million across the world in cellular network implementation.
Directed Researher
University of Southern California•  August 2022 - Present
Creating a new network technology surpassing 6G, being implemented on non-stationary BS (LEO SAT) by making use of Deep Reinforcement Learning coded on python, estimated to be a $1 Trillion+ industry soon. • Developing custom environments on VS Code, leveraging RLlib to interface environments with centralized and multi-agents for training across various Pytorch and TensorFlow based algorithms such as PPO, Impala and A2C.
Deep Learning intern
IEEE•  September 2020 - October 2020
• Fabricated a Brain Tumor Detector on MATLAB achieving an accuracy of 90%. Adapted segmentation and bone-masking techniques for detection, with the help of DL, ML & Statistical toolboxes.
Data Scientist (ML) intern
Tequed Labs•  January 2020 - February 2020
• Performed Data Analysis on jupyter notebook to develop a College Estimator for students. Executed model selection across 3 classification models i.e., Logistic regression, Decision-trees and KNNs from Scikit-learn. • Enhanced models with hyper parameter search and optimization getting a best accuracy of 93% among tested models.
Education
University of Southern California (USC), Los Angeles
Machine Learning and Data Science, MS