Piyush Raikwar
Doctoral Student at Faculty of Electrical Engineering, Czech Technical University in Prague
Geneva, Geneva, Czechia
Summary
🤩
Rockstar🎓
Top SchoolPiyush Raikwar is a doctoral student and machine learning engineer with eight years of experience building research-grade ML systems and high-performance numerical software. Based in Geneva and affiliated with Czech Technical University, he researches inductive biases in vision while previously developing fast calorimetry simulation with generative models at CERN. His open-source work includes a Bayesian CNN implementation in PyTorch focused on uncertainty estimation and a notable Google Summer of Code contribution to CuPy that added a NumPy fallback mode and ndarray wrappers for robust CPU/GPU interoperability. He has practical industry experience applying graph-based fraud detection at Swiggy and model compression, object detection, and federated learning in enterprise settings. Comfortable bridging research and production, he brings a rare combination of probabilistic deep learning expertise and low-level library engineering.
7 years of coding experience
4 years of employment as a software developer
Integrated Master's, Information Technology, Integrated Master's, Information Technology at ABV-Indian Institute of Information Technology and Management
Github Skills (12)
bayesian-network10
pytorch10
convolutional-neural-networks10
python10
cupy10
numpy10
testing9
machine-learning9
variational-inference9
data-transfer9
bayesian8
bayesian-inference8
Programming languages (5)
Github contributions (5)
kumar-shridhar/PyTorch-BayesianCNN
Feb 2020 - Aug 2020
Bayesian Convolutional Neural Network with Variational Inference based on Bayes by Backprop in PyTorch.
Role in this project:
ML Engineer
ML EngineerContributions:102 commits, 4 PRs, 35 pushes in 5 months
Contributions summary:Piyush primarily contributed to the development of a Bayesian Convolutional Neural Network in PyTorch. Their work involved implementing custom layers (BBBConv2d, BBBLinear), modifying the model architecture (BayesianLeNet, AlexNet, 3Conv3FC), and integrating uncertainty estimation techniques. The contributions are targeted to record the mean and variance of the Bayesian Neural Network.
bayesian-inferencebayesian-networkspymc3variational-inferencetemporal-data
cupy/cupy
May 2019 - Apr 2020
NumPy & SciPy for GPU
Role in this project:Back-end Developer
Back-end DeveloperContributions:195 commits, 7 PRs, 140 comments in 10 months
Contributions summary:Piyush appears to be focused on developing a fallback mechanism within the CuPy library. Their primary contribution involves implementing and refining a `fallback_mode` to use NumPy functions when corresponding CuPy functions are unavailable, indicating work on core functionality and library integration. They implemented a `RecursiveAttr` class to catch attributes and call relevant functions. This includes defining and integrating the `ndarray` wrapper. Furthermore, the user added the data transfer method between CPU and GPU and created tests to check compatibility with numpy functions and the handling of various cases for functions.
cudapythoncusolvergpunumpy
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.