Hello,

I'm Hema Deva Sagar, Potala

Machine Learning Engineer/Data Scientist

About me

Hello! My name is Hema Deva Sagar Potala. In short, you could refer me to as Sagar. Currently, I am pursuing my Master's in Computer Science at Texas A&M University.

My interest lay around the application of Machine Learning to solve problems.

During my 5 years of experience working as a Data Science Consultant at Deloitte Consulting, I worked on various problems like recommendation engines, cyber-attack detection, AI assited medical document reviewing, etc.

As part of my Master's, I started to focus on Computer Vision and application of Deep Learning on Visual Computing problems.

Work Experience

Novartis Pharmaceuticals - USA

Role: Data Science Intern

Duration: Jun'23 - Aug'23

Description: As part of my internship, I worked on the problem of recommending optimal advertising strategy that maximize drug sales, using deep learning.

Deloitte Consulting India Private Limited - India

Role: Consultant, Applied AI

Duration: June'17 - Dec'21

Description: As a data science consultant, I implemented machine learning solutions to various business problems. Some examples problems I worked on are recommendation engines, AI assited medical document reviewing, demand forecasting etc.

Texas A&M University, College Station - USA

Role: Graduate Assistant Teaching

Duration: Aug'22 - May'23

Description: As a teaching assistant, for data structure course, I assisted the course professor by conducting lab to the course students, grading, and holding office hours.

My Skills

Languages/Frameworks

C++

Python

Ruby

SQL

HTML

CSS

JS

Django

Rails

ML algorithms

Linear Regression

Logistic Regression

SVM

Random Forest

Gradient Boosting Machines

K-Means

PCA

R-CNN

ResNets

GAN

BeRT

NeRF

Projects

Generated portraits

Title: Semantic 3D-aware portrait synthesis and manipulation usng CNeRF

Explored Compositional Neural Radiance Fields, CNeRF, to generate portrait images and then manipulate various semantic regions within the generated portrait.

Title: Low Light Enhancement Using Zero-Reference Deep Curve Estimation

Built low light image enhancer by estimating image specific enhancement curves using a light weight deep neural network.

Title: CIFAR classification using NFNets

Traditional Batch-Normalization breaks independence between training examples, so, here I explored Normalization Free ResNets.

knowledge distillation

Title: Explore bias in knowledge distilled model

Explored the hypothesis that knowledge distilled smaller models will have biases amplified compared to the source model.

Publication

Singh S, Potala S, Mohanty AR. An improved method of detecting engine misfire by sound quality metrics of radiated sound. Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering. 2019;233(12):3112-3124. doi:10.1177/0954407018818693

Contact me

College Station, Texas, USA

hpotala@tamu.edu

+1 979 721-0431