Deepak Lingaraju

• Developed machine learning and hybrid deep learning models (CNNs, RNNs) for quality prediction and process analysis in resistance spot welding (RSW) using large-scale industrial sensor time-series data • Applied signal processing and feature engineering techniques (FFT, wavelet analysis, statistical feature extraction) to identify process patterns and build predictive models for nugget diameter estimation and weld quality assessment under varying process disturbances. • Implemented domain adaptation (e

• Built a custom convolutional backbone and integrated it into a YOLOv3–v4–inspired detection model tailored for garment region identification in automated processing • Designed a single-scale detection architecture optimized for fixed camera–garment distance and analyzed performance across multiple feature map depths for small vs. large objects. • Developed a custom YOLO-based loss function and conducted comparative experiments using transfer learning to benchmark model performance.