← Back to Students
Yael S.
Bio
Motivated and detail-oriented software developer with strong analytical and creative thinking. Fast learner with hands-on experience in software development, data workflows, and image processing.
Skills
Python
CUDA
PyTorch
OpenCV
GAN models
Stable Diffusion
Bootcamp Project
Framework for generating realistic SEM images from segmentation masks, sketches, or text
Mentored by: Applied Materials
Mentors:
Data Science Bootcamp 2025 (Data)
Responsibilities:
Research: Segmentation data augmentation for training – Implemented two masking approaches: training a U-Net model from scratch and applying OpenCV contour methods to identify layered regions.
Experimented with Stable Diffusion’s Textual Inversion by training custom embeddings in PyTorch on Kaggle GPUs to capture SEM-specific textures and patterns, evaluating their effect on improving realism in generated SEM images.
Trained a Pix2Pix model in PyTorch on Kaggle GPUs to convert Canny-based sketch inputs into realistic SEM images, including dataset construction, image preprocessing with OpenCV/NumPy, and model evaluation using FID, KID, SSIM and additional metrics.
Research: Add SEM noise to synthetic images – Added SEM-style noise by blending sampled noise patches from real SEM images and testing Gaussian/Poisson noise, with exploration of a future trained noise-injection model.
Click to enlarge
Additional Projects
- Recipe website - Developed a recipe-sharing website using React for the client interface and Node.js for the backend, integrating MongoDB Atlas for cloud-based data storage and Cloudinary for managing and storing user-uploaded images.
- Hospital shift management - .Net (server only)
English Level
Working Proficiency