Ezzaldeen Hamdan

Ezzaldeen Hamdan

AI engineer in Amman, Jordan

I'm an AI engineer based in Amman, Jordan, and an AI Fellow at Al Hussein Technical University (HTU). I work on how LLM models are adapted, combined and orchestrated, from letting one foundation model carry many specialized minds at once, to cohesive experiences across text, visuals, motion and video. Underneath it all is data: transforming vast, unruly collections into carefully curated corpora that models can genuinely learn from.

Download CVLinkedInGitHub
  • 1 of 20fellows selected for Jordan's national AI initiative
  • 3 h → 15 minduplicate-detection time cut at Altibbi
  • 2nd in cohortB.Sc. Computer Science, 3.69 GPA

In short

I'm one of 20 fellows in the inaugural cohort of 9XAI, Jordan's national AI initiative hosted by the National Council for Future Technology and Al Hussein Technical University, where we develop and validate AI solutions with the country's institutional and governmental partners.

Before that, I built a production medical RAG chatbot and a duplicate-detection engine at Altibbi, and loan-repayment prediction models at z28.ai. In my own time I build open-source tools for AI agents, such as House of Wisdom, an MCP server that consults several model families at once.

Role
AI Engineer
Based in
Amman, Jordan (UTC+3)
Current
9XAI Fellow (Lighthouse Cohort), since Feb 2026
Previously
z28.ai, Altibbi, Islamic International Arab Bank
Education
B.Sc. in Computer Science, Al-Ahliyya Amman University (2025)
Focus
LLM Post-Training, Parameter-Efficient Fine-Tuning, Data-Centric AI, Multi-Tenant LLM Serving, LLM Evaluation, Mechanistic Interpretability
Languages
Arabic (native), English (C2)
Open to
AI engineering and research roles, collaborations and interesting projects

Where I've worked

  1. 9XAI Fellow (Lighthouse Cohort)

    Al Hussein Technical University and National Council for Future Technology

    I was selected as 1 of 20 fellows for the inaugural national AI initiative, following a competitive, multi-stage evaluation process. The initiative, hosted by the National Council for Future Technology (NCFT) and Al Hussein Technical University (HTU), develops and validates AI solutions across Jordan's institutional and governmental partners while tackling nationwide challenges.

    • Applied AI
    • Project Management

  2. Artificial Intelligence Engineer

    z28.ai

    • Engineered memory-efficient ETL pipelines using Pandas chunking and BigQuery, processing multi-GB datasets (1M+ records).
    • Built an engine to identify and extract thousands of attributes from an unstructured HTML blob embedded in a database column.
    • Implemented automated data quality profiling, calculating null-rates and distribution statistics across dynamic schema columns.
    • Designed a snapshot modeling pipeline to generate training datasets for borrowers, loans, payments, and call-history features.
    • Conducted segment-level lift analysis to quantify repayment probability uplift across behavioral cohorts relative to baseline.
    • Implemented LightGBM models with class imbalance calibration, feature auditing, and cohort breakdowns, achieving 0.91 AUC.
    • Developed an ML interpretability pipeline to bridge the gap between high-performing ML models and business stakeholder trust.
    • BigQuery
    • Pandas
    • NumPy
    • LightGBM
    • Feature Engineering
    • Data Quality

  3. Artificial Intelligence Engineer

    Altibbi

    Duplicates Detector System

    • Built an end-to-end ETL pipeline for 10+ MySQL tables with data cleaning, content consolidation, and resumable processing.
    • Created a scalable embedding workflow with batching, caching, and storage of 100k+ vectors and metadata in ChromaDB.
    • Reduced the detection process from 3 hours to 15 minutes by building a rapid engine with cosine similarity and star clustering.
    • Added production safeguards, read-only DB access, metadata handling, a Streamlit UI, and config-based automated runs.

    Medical RAG Chatbot

    • Built an asynchronous web-scraping pipeline to collect and chunk 15k+ Altibbi news and articles with fault tolerance.
    • Designed and deployed a production embedding pipeline storing vectors in both Milvus and Qdrant for semantic search.
    • Implemented a 3-layer RAG stack (classification, refinement, retrieval) with OpenAI models and strict source citation.
    • Set up CI/CD pipelines with Jenkins and Bitbucket, integrating Docker-based builds to streamline deployment workflows.
    • NLP
    • LLMs
    • RAG
    • MLOps
    • Docker

  4. Rotational Intern

    Islamic International Arab Bank

    • Assisted in an Oracle 21c installation, learned Oracle RAC, and managed backups, LTO tapes, and card issuance.
    • Assisted in database administration and SDLC across 11 IT departments, focusing on middleware and API management.
    • Had exposure to enterprise banking systems (iMAL) and security frameworks, including ArcSight and intrusion prevention.
    • Oracle
    • SDLC
    • Database Administration
    • Security

Featured work

At Altibbi

Duplicate Detection Platform

An end-to-end system for finding near-duplicate medical content: a secure sync pipeline, a vector-based detection layer with cosine similarity and star clustering, and a Streamlit dashboard. It cut a 3-hour detection run to 15 minutes across 100k+ vectors.

  • Python
  • ChromaDB
  • Streamlit
  • Docker
Source on GitHub

At Altibbi

Medical RAG Chatbot

Answers Arabic medical questions from 15,000+ scraped Altibbi articles through a 3-layer RAG pipeline (classification, refinement, retrieval) with strict source citation. Vectors live in both Milvus and Qdrant, and builds ship through Jenkins and Docker.

  • Python
  • Milvus
  • Qdrant
  • OpenAI
  • Jenkins
Source on GitHub

Graduation project

Uni-Archives

An academic content-sharing platform that uses GPT models and Tesseract OCR to analyze uploaded course material, with integrated payments. Honoree graduation project at Al-Ahliyya Amman University.

  • C#
  • .NET Core
  • SQL Server
  • Bootstrap
Source on GitHub

University project

SecureScope

A machine-learning intrusion detection system that classifies live network traffic captured with PyShark using a LightGBM model.

  • Python
  • PyShark
  • LightGBM
Source on GitHub

Background & credentials

  1. 2021 – 2025

    B.Sc. in Computer Science

    Al-Ahliyya Amman University

    GPA 3.69/4.00 (Excellence), 2nd in cohort

    • Ranked 2nd in the Computer Science cohort, one of three students with an Excellence GPA.
    • Awarded a merit-based partial academic scholarship.
    • Honoree graduation project: Uni-Archives.
    • Member of the AAU IT and Data Science & AI clubs.

Certifications

  • Artificial Intelligence

    Samsung Innovation Campus

    AI program, The Levant 2025. Capstone project ranked 6th regionally with a score of 91.5/100.

  • Agentic AI Certificate

    SHAI

    Specialization in autonomous agent systems.

  • GitHub Foundations

    GitHub

    Git version control and collaboration on GitHub.

  • EF SET English Certificate

    EF SET

    C2 Proficient, the highest level.

Let's build something meaningful

Open to AI engineering and research roles, collaborations and interesting projects.

Send emailConnect on LinkedInView GitHub