Siddharth Gururani
Deep Learning Research Scientist at NVIDIA
California, United States
Summary
👤
Senior🎓
Top SchoolSiddharth Gururani is a deep learning research scientist at NVIDIA with 11 years of experience specializing in generative models for audio, speech synthesis, and music information retrieval. He earned a PhD in Music Technology from Georgia Tech where he developed weakly supervised models for multi-instrument recognition and contributed to feature learning and sparse coding research. Prior roles include AI Scientist at EA working on speech/voice technology and hands-on ASR engineering—he enhanced DeepSpeech2 training and logging in an open-source PyTorch repo. Siddharth bridges rigorous academic research with production-focused engineering, applying signal processing and representation learning to real-world audio systems. Based in California, he combines deep domain expertise in audio with practical experience shipping ML features at scale. An often-overlooked strength is his track record of improving tooling and reproducibility (e.g., checkpoint continuity and tensorboard logging) that makes experimental research more production-ready.
11 years of coding experience
6 years of employment as a software developer
Doctor of Philosophy (Ph.D.), Music Technology, 3.93, Doctor of Philosophy (Ph.D.), Music Technology, 3.93 at Georgia Institute of TechnologyB.Tech + M.Tech (Dual Degree), Computer Science and Engineering, B.Tech + M.Tech (Dual Degree), Computer Science and Engineering at Indian Institute of Technology, Kharagpur
English, French, Hindi
Github Skills (8)
tensorboard10
pytorch10
machine-learning10
speech-recognition10
python10
data-augmentation9
deep-learning9
nlp8
Programming languages (6)
Github contributions (5)
SeanNaren/deepspeech.pytorch
Jun 2017 - Aug 2017
Speech Recognition using DeepSpeech2.
Role in this project:
ML Engineer
ML EngineerContributions:6 commits, 3 PRs, 36 comments in 1 month
Contributions summary:Siddharth contributed to the training and logging functionality of a speech recognition model. They implemented tensorboard logging for tracking training metrics such as loss, WER, and CER, as well as logging model parameters. Additionally, they fixed issues related to continuing training from a checkpoint, ensuring the logging and visualization of previous training progress. The user also added sample rate configuration to the data augmenter.
deepspeech2recognitionspeechspeech-recognition
SiddGururani/sample_detection
Nov 2015 - Oct 2019
Contributions:15 pushes, 2 branches in 3 years 11 months
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.