Open to Junior Data Scientist roles

Mohamedalfateh Saeed

Data Scientist · Machine Learning & Computer Vision Researcher

I turn messy data into decisions — from exploratory analysis and statistical modeling to production-ready machine learning. PhD researcher with peer-reviewed IEEE & MDPI publications.

View Projects Get in touch

Peer-reviewed publications

27+

Datasets analyzed

14+

ML models built

99.9%

Accuracy on kidney CT

Selected work

Projects & case studies

Each project framed the way I work: a question, the data, the approach, and a measurable result.

TransConvNet hybrid Transformer-CNN architecture

TransConvNet — Hybrid Transformer-CNN for Medical Imaging

Lead Researcher · IEEE 2024

Question: Can kidney abnormalities be detected from CT scans accurately enough for clinical trust, where standard CNNs miss global context?
Approach: Built a hybrid model combining CNNs (local features) with Transformer attention (global context), plus Grad-CAM for interpretability.
Result: Reached 99.9% classification accuracy and produced visual explanations clinicians can inspect — a decision-support tool, not a black box.

PyTorchVision TransformersCNNGrad-CAMMedical Imaging

Read the paper

GenAI CRM & NLP Entity Extraction

Full-Stack · 2025

Question: How do you turn unstructured business-card images into clean, deduplicated, queryable leads in real time?
Approach: Gemini Flash 2.5 for entity extraction; SQLAlchemy + PostgreSQL schema tuned for deduplication; Flask REST endpoints.
Result: A live production CRM keeping the underlying dataset clean for downstream analysis.

Gemini Flash 2.5NLPFlaskPostgreSQL

Live demo

Facial emotion recognition with Grad-CAM heatmaps

Explainable AI — Facial Emotion Recognition

Researcher · 2022

Question: Which facial features actually drive a deep model's emotion predictions — and can we make that transparent?
Approach: CNN classifier on FER-2013 (~35k images) with Grad-CAM mapping predictions back to image regions.
Result: Heatmaps revealing the features behind each prediction, supporting bias checks and debugging.

TensorFlow/KerasGrad-CAMOpenCVXAI

Code

Scalable Image Classification API (MLOps)

ML Engineer · 2024

Question: How do you take a trained classifier from notebook to a reliable, reproducible service?
Approach: CNN behind a Flask API, Dockerized, served via Nginx/Gunicorn, with a CI/CD workflow to AWS EC2.
Result: An end-to-end deployment pipeline — the part of DS that turns a model into something usable.

DockerAWS EC2CI/CDFlask

Code

Tourism Data Analytics Dashboard

Data Scientist · 2023

Question: What do raw tourism reviews reveal about visitor trends, and how do you make that explorable?
Approach: Scraped and cleaned review data with Python/Pandas ETL, then built interactive visualizations.
Result: A dashboard turning messy reviews into at-a-glance, actionable insights.

PandasETLData VizReact

Code

End-to-End Tabular ML Coming soon

Data Scientist

Planned: A complete case study on a public tabular dataset: EDA, feature engineering, model comparison (scikit-learn), and honest evaluation — delivered as a documented Jupyter notebook.

PythonPandasscikit-learnJupyter

Where I've worked

Experience

Research Scholar

Apr 2025 – Present

TÜBİTAK & Osmaniye Korkut Ata University · Osmaniye, Türkiye

Investigated device heterogeneity in RSSI-based indoor positioning (TUJI1 dataset); built a CNN-LSTM framework reducing 3D positioning error to 2.20 m and improving cross-device robustness by 0.17 m — published in IEEE Access.
Built a hardware-agnostic CSI deep-learning framework (1D-ResNet + Squeeze-and-Excitation + Mixture-of-Experts with CORAL domain adaptation) for edge-deployable indoor localization, reaching 0.60 m mean error across heterogeneous IoT devices (manuscript under review).
Lead the ML component of a TÜBİTAK 1001 project on wireless node topology and live location detection under earthquake debris; designed algorithms improving detection accuracy by 2.4% (ongoing).

PythonPyTorch1D-ResNetMixture-of-ExpertsDomain AdaptationSignal Processing

PhD Research Intern (Erasmus+)

Aug 2025 – Oct 2025

AGH University of Science and Technology · Kraków, Poland

Conducted advanced statistical analysis and data curation for a study published in MDPI Electronics (2026) on the impact of environmental control on video Quality of Experience.

Statistical AnalysisData CurationScientific Reporting

Teaching Assistant

Nov 2017 – Feb 2025

Al Neelain University · Khartoum, Sudan (Hybrid)

Delivered 100+ online and onsite classes (35+ students each) across robotics, microcontrollers, and core engineering courses.
Supervised and evaluated 9 microcontroller projects for 80 students; co-developed course materials and curriculum.

TeachingMentorshipCurriculum

Teaching Assistant

Feb 2018 – Sep 2018

University of Medical Science & Technology · Khartoum, Sudan

Assisted lectures and guided 10+ students across 4 subjects and graduation projects; led 3 PLC/SCADA classes with 92.6% student success.

PLC/SCADAInstruction

Selective programs

Training & AI schools

ELLIS AI Winter School on Foundation Models

Mar 2026

ELLIS · Amsterdam, Netherlands

Studied LLM foundations, interpretability, post-training methods, and specialized models for tabular and multilingual data, plus the ethics of learning from human data.

Foundation ModelsLLMsInterpretability

M2L AI Summer School 2025

Sep 2025

University of Split · Split, Croatia · sponsored by Google DeepMind

Selected participant (18% acceptance, fully funded); 22h lectures + 10h workshops on generative/diffusion models, reinforcement learning, and LLMs.

PyTorchDiffusion ModelsRL

Software Engineering — Backend Specialization

Dec 2023 – Feb 2024

ALX Africa · Online · fully funded by Mastercard

Built RESTful APIs with Python (Django) and Node.js/Express; designed relational (MySQL) and NoSQL (MongoDB, Redis) stores; automated deployments with Docker and CI/CD.

DjangoNode.jsDockerCI/CD

Peer-reviewed

Publications

Hardware-Agnostic Deep Temporal Learning for Edge-Deployable Indoor Localization in Heterogeneous IoT Networks

TÜBİTAK Project · CSI Localization 0.60 m mean error 2026

Under review

Impact of Environmental Control on Subjective Video Quality Assessment in Crowdsourced QoE Experiments

MDPI Electronics (Journal) Impact Factor: 2.6 Apr 2026

View

Mitigating Device Heterogeneity for Enhanced Indoor Positioning System Performance Using Deep Feature Learning

IEEE Access (Journal) Impact Factor: 3.9 Oct 2025

View

TransConvNet: Enhancing Kidney Abnormality Detection in CT Imaging through a Hybrid Transformer-CNN Model with Integrated Explainability

IEEE COMSCI 2024 (Conference) Sozopol, Bulgaria 2024

View

Improvement of Human Activity Recognition (HAR) Performance by Utilizing LSTM in a Progressive Learning Structure

IEEE COMSCI 2024 (Conference) Bulgaria Sep 2024

View

Toolbox

Skills

Languages & Querying

PythonSQLBashC

Analysis & Statistics

PandasNumPyMatplotlibSeabornEDAHypothesis TestingFeature Engineering

Machine Learning

scikit-learnTensorFlow/KerasPyTorchModel EvaluationDeep LearningComputer VisionNLP / LLMsGrad-CAM (XAI)

Tools & Workflow

Git & GitHubJupyterDockerAWSPostgreSQLLinux

Also familiar with: Flask, Django, Node.js, MongoDB, Redis, Nginx, CI/CD — from production ML & backend projects.

Academic background

Education

Ph.D.

Electrical & Electronics Engineering

Ondokuz Mayis University · Samsun, Türkiye

Sep 2024 – PresentGPA 3.93/4.0Gov. Scholarship

M.Sc.

Electrical & Electronics Engineering

Istanbul University-Cerrahpaşa · Istanbul, Türkiye

Sep 2019 – Jun 2022GPA 3.38/4.0Gov. Scholarship

B.Sc.

Electronics Engineering (Control)

Al Neelain University · Khartoum, Sudan

Sep 2012 – Oct 2017Top of class · 83.67%Best Project Award

Continuous learning

Certifications & courses

Machine Learning Specialization

Stanford (Coursera)

May 2023

Deep Learning Specialization

Stanford (Coursera)

Dec 2023

Generative AI with LLMs

DeepLearning.AI

Jan 2026

AI Agents & Agentic AI in Python

Vanderbilt (Coursera)

Mar 2026

Retrieval Augmented Generation

Stanford (Coursera)

Ongoing

About me

Engineer turned data scientist

I started in control engineering and moved into machine learning and computer vision. I work end to end: framing the question, cleaning and exploring data, building and validating models, and communicating results clearly enough to drive a decision.

My research has produced peer-reviewed IEEE and MDPI publications, and I care as much about rigor and reproducibility as I do about results. Originally from Sudan, currently based in Samsun, Türkiye — open to relocation and remote roles.

Let's build something with data

Open to junior / applied data scientist roles and research collaborations. I usually reply within a day.

Email me LinkedIn GitHub