Samuele Cornell
Software Engineer at Carnegie Mellon University
Pittsburgh, Pennsylvania, United States
Summary
🤩
RockstarSamuele Cornell is a software engineer with a decade of experience specializing in speech and audio ML, currently a Post-Doctoral Research Associate at CMU LTI based in Pittsburgh. He has hands-on expertise implementing and refining core models and tooling in prominent open-source projects such as SpeechBrain, ESPnet, and Asteroid, contributing VAD examples, DPRNN/DPTNet architectures, and learnable Fourier positional encodings. Comfortable across research and engineering boundaries, he pairs model design and training skills with pragmatic back-end development, linters/formatting, and bug fixes that improve code quality and reproducibility. His work shows a practical focus on loss formulations and architectural tweaks that enable cleaner training and future-proof extensibility. Colleagues benefit from his blend of research rigor and production-minded implementation, especially in complex audio-source-separation and speech-processing stacks.
10 years of coding experience
Stackoverflow
165reputation
34kreached
5answers
3questions
Github Skills (21)
pytorch10
python10
machine-learning10
deep-learning10
transformer10
speech-recognition10
speechbrain10
machine-learning-models9
n9
rnn-model9
audio-processing9
deeplearning-ai9
mlops8
image-preprocessing6
pca6
Programming languages (6)
Github contributions (5)
speechbrain/speechbrain
May 2020 - Jan 2023
A PyTorch-based Speech Toolkit
Role in this project:
ML Engineer
ML EngineerContributions:23 reviews, 364 commits, 10 PRs in 2 years 8 months
Contributions summary:Samuele's commits primarily involve developing and refining a Voice Activity Detection (VAD) minimal example within a PyTorch-based speech processing toolkit. The contributions include the initial implementation of the VAD example, integrating binary cross-entropy loss functions, refining the script, and adding a permutation-invariant loss wrapper for potential future applications. The user demonstrated skills in model training, feature extraction, loss function implementation, and code refinement within the speech processing domain.
voice-recognitionasrspeech-recognitionspeech-separationspeaker-verification
asteroid-team/asteroid
Nov 2019 - Mar 2021
The PyTorch-based audio source separation toolkit for researchers
Role in this project:ML Engineer
ML EngineerContributions:51 reviews, 66 commits, 39 PRs in 1 year 4 months
Contributions summary:Samuele's commits primarily focused on implementing and refining the Dual-path RNN (DPRNN) architecture within the `asteroid` repository. They added new functionalities for Dual Path Transformer Network (DPTNet), and also modified various aspects of the code including adding batch normalization and fixing various bugs. The user made changes across multiple files, indicating hands-on involvement in the core model design and implementation of machine learning components for audio source separation.
pytorchsource-separationdeep-learningaudio-separationaudio-source-separation
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.