Shahd Mohamed

I’m an AI Engineer with hands-on experience in Machine Learning and Deep Learning, working across NLP, Generative AI, and Computer Vision. I’m especially driven by my growing interest in Explainable AI to ensure models are reliable and safe.

In my projects, I focus on making AI both effective and trustworthy. I led Knowledge Distillation of BioViL-T for chest X-ray image-to-text retrieval, including semantic-aware hard negative fine-tuning and an ablation study—achieving a best student at 383% of teacher R@1. I also built an XAI-driven lymphoma diagnosis workflow using fine-tuned ResNet and DenseNet, applying GradCAM++ to verify predictions rely on clinically relevant structures rather than artifacts.

I’ve also developed real-world AI prototypes and systems that integrate modern LLM tooling and deployment-ready stacks. I created an adaptive learning platform that generates quizzes and feedback from topics, documents, or past mistakes using the Azure OpenAI API and LangChain, and a real-time review summarizer using ABSA with BERT/RoBERTa/DistilBERT plus a Gemini API. Through roles like my AI Intern experience at Samsung Innovation Campus and my Data Engineering Trainee work at the Digital Egypt Pioneers Initiative, I strengthened my ability to move from training and experimentation to structured data and practical solutions.