A picture of Swetha

Swetha Mandava

About me. I'm currently working on building You.com and experimenting with radically new ways to help our users confidently make better decisions. If you're interested in becoming a beta user, hit me up! Previously, as a Senior Deep Learning Engineer at Nvidia, I worked on developing deep learning algorithms, especially NLP models and scaled them for large GPU clusters.

I graduated from Carnegie Mellon University with a Master's in Electrical and Computer Engineering and Bachelor's in Electronics and Communication Engineering from Manipal Institute of Technology. While at CMU, I was advised by Sebastian Scherer at The Air Lab where I worked on coverage planning for aerial robots. I was also advised by Gauri Josh while I researched hyperparameter tuning methods.

Research interests. I'm interested in efficiency - how can we do better and faster with lesser? To this end, I'm interested in designing and optimizing Deep Learning algorithms. I'm also intrigued by stories and people - how can we shape algorithms to make us better, happier and most importantly, kinder!

If you'd like to collaborate on research, chat with me about AI/startups or network with me - you can book time on my calendar here or drop a note below..


Articles
Talks
Misc/Media
Becoming Data-Centric at You.com - a privacy focused search engine Swetha Mandava, Zairah Mustahsan Neural Information Processing Systems 2022
Algorithmic and Software Techniques to Optimize BERT Training and Inference Swetha Mandava, Sharath Turuvekere Sreenivas GPU Technology Conference 2021
Pay Attention When Required Swetha Mandava, Szymon Migacz, Alex Fit-Florea The Batch Newsletter Coverage
Distributed Large Batch Training Swetha Mandava Deep Learning for Science School 2020
GANDALF: Generative Adversarial Networks with Discriminator-Adaptive Loss Fine-tuning for Alzheimer’s Disease Diagnosis from MRI Hoo-Chang Shin, Alvin Ihsani, Ziyue Xu, Swetha Mandava, Sharath Turuvekere Sreenivas, Christopher Forster, Jiook Cha, Alzheimer’s Disease Neuroimaging Initiative International Conference on Medical Image Computing and Computer-Assisted Intervention
Prototype to Production: How to Scale your Deep Learning Model. Swetha Mandava, Alex Qi Grace Hopper Conference 2019
Pretraining BERT with Layer-wise Adaptive Learning Rates Christopher Forster; Thor Johnsen; Swetha Mandava; Sharath Turuvekere Sreenivas; Deyu Fu; Julie Bernauer; Allison Gray; Sharan Chetlur, Raul Puri
BERT Meets GPUs Sharath Sreenivas, Swetha Mandava, Boris Ginsburg and Chris Forster