You can deploy this model and start earning money today!

Skully01/
Skully01

Views

No views yet

Hi, I'm Ahmed Youssef 👋

An AI & Machine Learning Engineer Passionate About Building Intelligent Systems

binarskull

I'm an AI/ML Engineer with a Bachelor's in Computer Science, specializing in Intelligent Systems. My passion lies in transforming complex, real-world problems into high-impact solutions through code and manipulating AI systems creatively to get the best results. I have hands-on experience developing LLM-based applications, implementing computer vision models for medical imaging, and building end-to-end data analysis pipelines.

🔭 I’m currently diving deep into Agentic AI and trying to learn advanced systems with the Hugging Face AI Agents Course.
🌱 My goal is to continuously bridge the gap between cutting-edge research and practical, deployable AI solutions.
💬 Ask me about PyTorch, Computer Vision (CV), Large Language Models (LLMs), and MLOps.
📫 Let's connect! Reach out to me on my LinkedIn or shoot me an email. ↙↙🔫

Gmail

Portfolio

🛠️ My Tech Stack

My toolbox is equipped with the latest technologies for building robust and scalable AI systems from the ground up.

Core AI/ML	Python PyTorch TensorFlow Scikit-learn Pandas LLMs
Tools & Platforms	AWS Docker Git SQL FastAPI
Concepts & Methods	RAG Prompt Engineering MLOps Agile

🚀 Featured Projects

Here are some of the projects I'm most proud of. They showcase my skills in deep learning, MLOps, and data-driven problem-solving.

Project	Description	Tech Stack
🧠 AI-Powered Glioma Segmentation	My graduation project, where I adapted the BEFUnet (CNN-Transformer) architecture to segment gliomas from multimodal MRI scans. Achieved a mean IoU of 0.805 on the BraTS 2020 dataset by building a full end-to-end training and evaluation pipeline.
🏭 End-to-End Landmark Classifier	Designed a complete MLOps-lite workflow to classify landmark images. This project included data processing, comparing a custom CNN vs. a fine-tuned ResNet18, implementing `pytest` unit tests, and packaging models with `torch.jit` for deployment and using FastAPI and Docker to containerize and use the classification endpoint.
⚡ Model Optimization Analysis	Executed a comparative study on a ResNet50 model to evaluate optimization techniques. Achieved up to 90% model size reduction (245MB to 23MB) via quantization and a 2x inference speed-up through knowledge distillation, all while managing accuracy trade-offs.
🚲 Bike Demand Prediction on AWS	Developed a predictive model for a Kaggle competition using AWS SageMaker Studio. This project leveraged AutoGluon with custom hyperparameter optimization (HPO) to forecast bike sharing demand, transitioning from descriptive analysis to building a cloud-based ML solution.