Kushal Silwal

• Developed a comprehensive semantic segmentation pipeline for microstructure analysis of metallographic images. • Designed and trained a robust U-Net segmentation model using PyTorch, enhancing characteristics detection accuracy. • Configured high-performance GPU training on an HPC cluster, automating the training pipeline with SLURM for efficiency. • Created visualization tools for model assessment, including IoU and Dice coefficient calculations for performance tracking.

• Conducted quality control for exhaust gas measuring devices, ensuring compliance with industry standards. • Diagnosed errors in PN meters, facilitating timely maintenance and creating comprehensive technical documentation. • Utilized Python scripts for software troubleshooting and adjustments, enhancing device performance. • Set up PN meters in the laboratory for calibration, contributing to accurate measurement processes.

• Assisted in the installation, maintenance, and troubleshooting of industrial machinery and automation systems. • Gained hands-on experience in metalworking, including machining, joining and materials processing techniques. • Conducted quality assurance for springs through fabrication, microscopic analysis, and stress testing.