Neha Upadhyay

About Me

Hello World! I'm Neha. I hail from a pure Computer Science background and you can tell that about me by my obsessive use of binary logic in everyday life. Jokes apart, I'm a researcher in the field of medical image analysis because I love to segment and analyze MRI, CT, and all kinds of scanned images of the brain. I develop deep learning models to achieve better performance and higher segmentation accuracy. Basically, my Neural Nets emulate the functioning of the human brain in order to analyze a scan of the actual brain so that we can better understand and identify issues with it.

I'm currently a Software Engineer at Envestnet Yodlee, Bengaluru. As a part of the Site Reliability Engineering team, my day-to-day professional life includes developing monitoring dashboards, enabling synthetic monitoring, writing scripts in python and shell, and analyzing as well as resolving incidents within the SLA.

Apart from my academic and professional life, I am passionate about sustainability and I run my own business venture, Gaia. At Gaia, we curate sustainable, ethically-sourced, organically-grown products from local artisans and small businesses into handmade boxes of joy we like to call “Eco-friendly Hampers.” Developing & maintaining the website, managing the operations & delivery, sourcing & curating the products, creating & posting the blog content, styling & marketing the products for Gaia are just some of the things that keep me going every day!

Experience

Envestnet Yodlee

Software Engineer

Wrote extensive multi-threaded JDBC programs for the Oracle database to PostgreSQL in AWS migration project for execution of 7000+ SQL queries in less than 30 minutes.
Designed end-to-end automated solutions, curl, telnet, synthetic monitoring scripts in python and shellscript and php dashboards to bring down the number of incidents missed by 60%.

National Institute of Technology Durgapur

Research Assistant

Designed an Energy Based Generative Adversarial Network that exhibited higher stability than normal GANs and a dice coefficient of 88.7.
Developed CSNet: A composite deep neural network for segmentation that topped the leader boards in the MICCAI ISLES 2015 challenge.

MAQ Software

Software Development Intern

Formulated compact code to integrate two intermediate databases into one to improve the efficiency of dependent analytical processing cubes.
Worked with technologies like SQL Server Management Studio, Integration Services, Microsoft Power BI, Visual Studio among other tools.

Indian Institute of Technology Delhi

Research Intern

Designed a tracking engine for the architectural simulator, TEJAS, with the aim of extending it to simulate Android architecture in addition to the x86 family.

Projects

E-commerce Website for a Sustainable living brand, Gaia

Used Wordpress, Woocommerce and CSS to build an e-commerce website for a self-owned sustainable living brand, Gaia. Gaia is a green venture that offers eco-friendly gift hampers comprising of local and biodegradable products curated from all over India.

CSNet: A New DeepNet Framework for Ischemic Stroke Lesion Segmentation

Built a composite deep neural network conflating fractal networks and the U-Net model for state-of-the-art segmentation results. Publication: CSNet: A New DeepNet Framework for Ischemic Stroke Lesion Segmentation, Computer Methods and Programs in Biomedicine, Elsevier.

View More Projects

Energy-Based Generative Adversarial Network (EBGAN) for Effective Semantic Segmentation of Brain Gliomas

Designed an EBGAN model exhibiting higher stability than regular GANs and demonstrating better segmentation performance.EBGAN employs an energy function to discriminate between the real and synthetically generated samples, analogous to probabilistic GANs

PIRNet:Two-step Deep Neural Network for Segmentation of Brain MRI with Efficient Loss Functions

PIRNet, based on a novel technique called Pyramid Pooling, together with the combination of inception blocks and residual connections, has been applied to the task of brain MRI segmentation of cerebrospinal fluid, gray matter, white matter, and background over several loss functions.