Saswat Das
Joint Secretary at Data Science Group, IIT Roorkee
Uttarakhand, India
Summary
🤩
Rockstar🎓
Top SchoolSaswat Das is a technical lead with 9 years of experience, currently guiding the Runtime team at Postman and tackling unconventional HTTP use cases daily. With a strong foundation from IIT Roorkee in applied mathematics, he blends rigorous academic training with hands-on ML and backend engineering across open-source ecosystems. His contributions span notable projects like PyTorch Vision and CuPy—where he improved test automation, dataset support, and sparse linear algebra functionality—reflecting a focus on reliability, performance, and reproducible research. He has research experience from IISc in adversarial robustness and unsupervised domain adaptation, demonstrating an appetite for hard problems at the intersection of vision and robustness. Based in Uttarakhand, he balances engineering with curiosity-driven hobbies—experimenting with new tools and building robots—bringing a pragmatic yet inventive approach to system design and developer experience.
9 years of coding experience
DAV Public School, Pokhariput , BhubaneswarIndian Institute of Technology Roorkee
French, Odia, Hindi, English
Github Skills (26)
pytorch10
scipy10
pytest10
python10
data-science10
sparse-matrix10
machine-learning10
numpy10
zoo10
mask-rcnn10
deep-learning10
cupy10
flux10
computer-vision10
faster-rcnn10
Programming languages (8)
Github contributions (5)
cupy/cupy
Mar 2020 - Jan 2022
NumPy & SciPy for GPU
Role in this project:
Back-end Developer
Back-end DeveloperContributions:3 reviews, 16 commits, 2 PRs in 1 year 10 months
Contributions summary:Saswat contributed to the implementation of sparse matrix norm calculations within the CuPy library, focusing on supporting various norm types for sparse matrices. Their work involved adding a new function `norm` and modifying existing files related to sparse linear algebra. The user also incorporated and tested the changes, extending the functionality of CuPy's sparse matrix capabilities. Further commits involved merging and integrating these updates into the main branch of the repository.
cudapythoncusolvergpunumpy
dsgiitr/d2l-pytorch
May 2019 - Aug 2019
This project reproduces the book Dive Into Deep Learning (https://d2l.ai/), adapting the code from MXNet into PyTorch.
Role in this project:Data Scientist
Data ScientistContributions:7 commits, 11 PRs, 10 comments in 3 months
Contributions summary:Saswat contributed to the development and improvement of deep learning models within the repository. Their work included modifying existing convolutional neural networks (LeNet) and implementing a new Network in Network (NiN) model. Furthermore, they enhanced the training process by optimizing loss and accuracy calculations using torch tensors and vectorization, demonstrating a focus on performance improvements. They also provided detailed descriptions of changes and improvements made.
d2lnlppytorchmxnetpytorch-implmention
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.