Junior AI Engineer | LLMs, RAG & Prompt Engineering | Building Practical AI Systems
Omar is a Generative AI & Machine Learning Engineer with practical experience across the full ML lifecycle, including LLMs, RAG, Prompt Engineering, Deep Learning, and Computer Vision. He focuses on building scalable, production-ready AI systems that automate workflows and solve real-world problems.
Expertise in designing and implementing LLM-backed systems, RAG pipelines, and prompt engineering to extract and summarize unstructured text, ensuring consistent and contextually accurate outputs.
Strong foundation in data analytics, ML engineering, and end-to-end pipeline development from data ingestion and feature engineering to model deployment. Experienced with CNNs, regression, and classification models.
Focus on building reliable, production-grade AI software, emphasizing architectural decisions, defining non-functional requirements, and implementing robust testing for AI pipelines to ensure accuracy and prevent hallucinations.
Built a production-grade RAG pipeline that ingests course material, chunks and embeds content into Qdrant, and generates contextually accurate exam questions via LLM.
Impact: Reduced average generation latency by ~60% under concurrent load by parallelizing LLM calls.
Analyzed student lifestyle and academic behavior data, identified key performance drivers through EDA, and built regression and classification models to predict exam scores.
Impact: Packaged models in an interactive Streamlit dashboard for real-time KPI monitoring.
Scraped and cleaned structured market data from online listings and built regression models to predict resale prices for the Egypt Used-Car Market.
Impact: Automated price-tracking tasks with scheduled Python scripts, improving monitoring efficiency by 40%.
Trained and benchmarked RNN, LSTM, and BiLSTM architectures on the IMDB dataset for sentiment classification.
Impact: Achieved 89% classification accuracy with the BiLSTM model.
Led data analysis of perception system outputs and built visual dashboards to track detection accuracy trends for the DU Racing Team.
Designed structured prompts with LangChain templates to extract and summarize unstructured text, enforcing consistent JSON output schemas.
Engineered automated feature engineering pipelines and trained CNN-based image classifiers, achieving 89% accuracy as a Data & Machine Learning Intern.
Finalist in the GenAI Agent Hackathon.
Expected to graduate with a B.Sc. in Computer Science & Artificial Intelligence from Damietta University.