Salalah, Oman — Data Scientist & AI Engineer

Models are the easy part. Getting them into production is the work.

I build machine learning and computer vision solutions — and the systems that actually run them. Right now that means AI and data work for two contracting groups in Oman, while finishing a master's in data science at Cairo University.

Download CV (PDF) See the work
A regression plot: a fitted line through scattered points, with the residual gap from each point to the line drawn in. the model the work
Fig. 1 — The line is the model. The gaps are the job.

MSc Data Science — Cairo University, in progress BSc Engineering — Zewail City Python · TensorFlow · Spark · Odoo

Now

Two groups, one brief: make the data useful.

2026 — Present · Salalah, Oman

Salem Mohiyaddin Bin Saif & Bros.

AI and data science across a group founded in 1975 that now spans twelve companies. I built the group website, and I maintain and extend the Odoo ERP and the data flows around it — turning operational records into something a manager can act on.

salemgroups.com
2026 — Present · Salalah, Oman

Al-Hashemi & Al-Rawas

The group's civil engineering and contracting arm — around a thousand employees. The same brief here: their Odoo ERP, the website I built and still run, the company data, and the AI solutions on top of it.

alhashemigroups.com
Selected work

Deep learning, computer vision, and data at scale.

Sentiment analysis with LSTMs

Deep learning · NLP · Python, TensorFlow, NLTK

83% test accuracy on 1.6M tweets — above most published results on the same dataset

Trained LSTM networks on a 1.6-million-tweet Kaggle corpus, with text preprocessing in NLTK and regex and data preparation in pandas. A transformer variant reached 80%, which made the comparison between architectures the interesting part of the project rather than the headline number.

  • TensorFlow
  • LSTM
  • Transformers
  • NLTK
  • pandas

Car plate detection & OCR

Computer vision internship · Zewail City · 2022

Detection with YOLO and Detectron2, then OCR for character extraction

A two-stage pipeline: locate the plate in the frame, then read it. Built robust detection models across both frameworks and implemented OCR for accurate plate-number extraction — the kind of problem where the model is fine and the edge cases are everything.

  • YOLO
  • Detectron2
  • OpenCV
  • OCR

200,000 matches, analysed in Spark

Big data · Python, Spark, Hadoop, Tableau

15+ map-reduce jobs over 200k matches pulled from the Riot API

Retrieved match data at scale, ran more than fifteen map-reduce jobs with Spark to find what actually correlates with winning, and presented the findings to stakeholders in Tableau. Repeated the same pipeline on a Goodreads dataset to check the approach generalised.

  • Apache Spark
  • Hadoop
  • Tableau
  • ETL

Image search & retrieval engine

Computer vision · OpenCV, scikit-image, Flask

Two retrieval modes: CBIR on global features, and SIFT descriptor matching

A web application that finds images by content rather than filename. Feature extraction runs on the server, with a Flask front end so the whole thing works as an actual searchable tool instead of a notebook.

  • CBIR
  • SIFT
  • OpenCV
  • Flask

Image classification with bag of words

Classical ML · scikit-learn, OpenCV

Feature extraction → visual bag of words → LSVC and KNN compared

Built the classical pipeline end to end and put two classifiers head to head. Worth doing before reaching for deep learning: it makes clear what the network is actually buying you.

  • scikit-learn
  • LSVC
  • KNN
  • BoW
Technical skills

What I work with.

Machine learning

Models

  • TensorFlow · PyTorch
  • LSTM · Transformers
  • SVM · KNN · scikit-learn
  • YOLO · Detectron2
Data

Scale & pipelines

  • Apache Spark · Hadoop
  • pandas · NumPy · SQL
  • ETL & data cleaning
  • Tableau · Matplotlib
Engineering

Shipping it

  • Python · C/C++ · C#
  • Flask · REST APIs
  • Odoo ERP · PostgreSQL
  • Git · OOP · algorithms
Foundations

The theory

  • Statistical inference
  • Regression & estimation
  • Linear & nonlinear optimization
  • NLP · speech recognition
Education

Statistics in Cairo, engineering at Zewail City.

Spring 2026 — expected 2028

MSc, Data Science

Faculty of Graduate Studies for Statistical Research, Cairo University — thesis in progress.

2016 — 2023

BSc, Communications & Information Engineering

University of Science and Technology at Zewail City. Deep learning, computer vision, NLP, speech recognition, big data analytics, statistical inference, and optimization.

Certificates: Introduction to Data Science, Networking Basics and Introduction to Cybersecurity (Cisco Networking Academy) · Introduction to Python (DataCamp)

Moaaz Al-Akel

Salalah, Dhofar

About

From robotics competitions to production ML.

I started out coaching the Dhofar team for a national robotics competition, went to Zewail City to study communications and information engineering — organising Egypt's VEX Robotics National Championship along the way — and came out the other side doing computer vision and big data. The through-line has always been the same: take something messy and make it work reliably.

I did freelance data analysis on Upwork for clients worldwide, mostly big-data work in Spark and Hadoop. Now I'm applying the same toolkit inside two Omani contracting groups, where the datasets are less tidy than Kaggle's and the results have to hold up in front of people who run the business.

I work in Arabic and English.

Contact

Open to data science and AI roles.