SOURABH YADAV

Dynamic Formal Verification and Machine Learning Engineer with a proven track record at NVIDIA, specializing in formal verification methodologies and machine learning. Pioneered LLM-based tools that significantly reduced testbench setup time. Proficient in Python and adept at cross-functional collaboration, driving innovation in verification workflows and enhancing simulation performance.

* Super-Resolution U-Net: Built dual-level residual U-Net achieving SSIM of 0.97 on validation set, outperforming existing models.

* Multilingual Emotion Detection: Designed parallel 1D CNNs for spectrogram + NLP-based emotion classification across 10 classes.

* Sepsis Predictor: Developed edge-compatible deep learning model predicting sepsis 6 hours prior to onset, deployed via Flask web & Android app.

• Multilingual Emotion Analysis from Speech - Parallel 1D CNN for emotion classification. Link:- https://link.springer.com/chapter/10.1007/978-981-19-3679-1_36
• A model for teaching geometry to schoolchildren using smartphone app - Analyzed smartphone-based pedagogy for school students. Link:- https://ieeexplore.ieee.org/abstract/document/9768570
• Smartphone Data Analysis for Nomophobic Behavior Among University Students During Lockdown in India - Studied behavioral data from university students using smartphone sensors. LInk:- https://ieeexplore.ieee.org/abstract/document/10463251

Tutorial Author: ML/DL content on Kaggle - https://www.kaggle.com/sourabhy