profile = {
"name": "Ibrahim Al-Shawa",
"role": "AI Engineer",
"focus": ["LLM Systems", "ML / Deep Learning", "Prompt Engineering"],
"writing": "Technical Writer — system design docs, SOPs, prompt specifications",
"leadership": "Technical Head @ Enactus Port Said (Aug 2025 – present)",
"education": "B.Eng. — Computer & Control Engineering, Port Said University, Egypt",
"research": "PBTune — Population-Based Training for DB parameter tuning [open-source]",
"interests": ["Agentic AI", "Production ML Systems", "AI Research", "Software Architecture"],
}I sit at the intersection of language, intelligence, and systems — engineering AI solutions end-to-end, from prompt design to deployed pipelines, while writing the documentation that makes them maintainable.
|
AI / LLM Engineering
|
Machine Learning & Deep Learning
|
Engineering & Architecture
|
The first application of Population-Based Training (DeepMind's evolutionary hyperparameter-optimization algorithm) to automatic database parameter tuning. PBTune maintains a population of configurations that evolve online — poor performers exploit elite configs and explore nearby variations — discovering high-performance database settings without a DBA and without any pre-training phase.
Scores competitive performance metrics (latency / throughput), in 5x the wall-clock seconds of SOTA tuning tools.
Python·PostgreSQL·Docker·GCP Compute Engine·Evolutionary Optimization·HPO·Research
A turnkey Software Development Standard Operating Procedure for driving agentic AI development from a Product Requirements Document (PRD) all the way to production deployment — with no gaps. Leverages a curated stack of prompting techniques.
Built to be the definitive guide for anyone using AI IDEs or agents to ship real software.
Prompt Chaining·ReAct·Contextual Priming·Negative Prompting·Tree of Thought·Self-Consistency
A living "documenting-as-learning" reference — a structured guide covering ML and Deep Learning from first principles to expert-level topics, with practical coding examples and clear learning paths organized by interest area.
If you're learning ML/DL and want a single well-organized, practitioner-first resource — this is for you.
Machine Learning·Deep Learning·Jupyter·Python·Educational
A multi-modal ML system for predicting cancer-related diseases using:
- Tabular data pipeline — classical ML with feature engineering & ensemble methods
- Image recognition pipeline — CNNs for histopathological image classification
scikit-learn·TensorFlow / Keras·Pandas·NumPy
A feature-complete desktop audio player built using the Qt framework in Python, developed before AI-assisted development became mainstream — written by hand, every line.
Python·PyQt·Qt Framework·Desktop App
| Certification | Issuer | Status |
|---|---|---|
| 🏆 Machine Learning Specialization | DeepLearning.AI / Coursera (Andrew Ng) | ✅ Completed |
| 🏆 Deep Learning Specialization | DeepLearning.AI / Coursera (Andrew Ng) | ✅ Completed |
Technical Committee Head — Enactus Port Said · Aug 2025 – Present
Enactus is a global non-profit operating in 36 countries, uniting student, academic, and business leaders to develop social entrepreneurship projects with real-world impact. As Technical Committee Head of the Port Said branch, I oversee the technical strategy, tooling, and execution of the committee's engineering and AI initiatives.
