Youssef ElhelwAI

I build practical AI applications across machine learning, computer vision, NLP, RAG, and large language models, with a focus on understanding model fundamentals and turning them into reliable real-world solutions.

During my AI internship at Edrak-Tips Hindawi, I worked with Transformers, LLMs, and Retrieval-Augmented Generation. I built a YouTube video summarization application and an AI system that extracts and parses structured information from CVs.

I developed P.O.W., an AI-powered planning assistant for team leads that generates project workflows, file structures, and skill-based task assignments. Its three-stage LLM pipeline uses LangChain and Qwen2.5-7B-Instruct, with Streamlit, FastAPI, Hugging Face Transformers, Pydantic, Kaggle, and ngrok supporting the end-to-end application.

I also built a movie recommendation web app using TF-IDF, cosine similarity, content-based filtering, feature engineering, and web scraping to recommend ten movies with similar “vibes.”