Gebrearegawi Gebremariam Gidey is an esteemed instructor and researcher in the Faculty of Computing Technology at Aksum University, Ethiopia. He is passionately committed to advancing artificial intelligence (AI), machine learning (ML), and natural language processing (NLP), particularly for low-resource and underrepresented languages.

He holds a Master of Science in Information Technology, specializing in Machine Learning and Artificial Intelligence, and graduated with a Bachelor of Science in Information Technology from the University of Gondar as an Academic Gold Winner. Additionally, in 2025, he will earn a Bachelor of Theology (B.Th.) from Saint Freminatos Abba Selama Theological College, enhancing his technical background with insights into Ge'ez and cultural heritage.

His innovative research focuses on developing AI solutions for complex languages in the Horn of Africa, including Ge'ez and Tigrinya. He has published works such as a Morphological Synthesizer for Ge'ez and a study on English-Tigrinya Machine Translation, and his research also extends to database security and cloud adoption for education. His interdisciplinary interests combine digital technologies with the preservation and analysis of ancient Geʽez manuscripts.

Since 2016, he has taught undergraduate courses, supervised theses, and led community technology projects. Currently, he is the Postgraduate Coordinator and a member of the Academic Council at Aksum Institute of Technology.

His research focuses on AI, NLP, large language models (LLMs), and big data. He is dedicated to digital inclusion and cultural preservation, empowering communities through innovative technology.

CV (PDF, 633 KB)

• Natural Language Processing for low-resource and ancient languages (especially Ge'ez)
• Morphological analysis and synthesis of Semitic languages
• Digital preservation of Ethiopian manuscript heritage
• Artificial Intelligence and Machine Learning applications for cultural heritage preservation

Current research project

Digital Preservation of Ge'ez Manuscripts: Handwritten Ge’ez Script Recognition in Ancient Ethiopian Manuscripts Using DCNN

The ancient Ge'ez script underpins Ethiopia's rich literary and religious heritage, preserved in thousands of fragile manuscripts threatened by physical degradation and structural complexity. Existing OCR efforts for Ge'ez remain limited by incomplete character coverage, insufficiently diverse datasets, and poor handling of historical degradation.

This study addresses these gaps by developing an end-to-end Deep Convolutional Neural Network (DCNN) framework for accurate recognition of handwritten Geʽez scripts. We will create a robust, expert-validated dataset of 30,000–40,000 ancient written character images, annotated to Unicode standards and augmented to simulate real manuscript degradation. A custom DCNN architecture will be designed, trained, and benchmarked against transfer learning models (VGG–16 and ResNet–34), to achieve high accuracy even on severely degraded documents.

Ultimately, this research delivers an effective digital preservation tool, enabling broader scholarly access to Ethiopia's invaluable historical records while safeguarding this vital cultural heritage for future generations.

• Gidey, G. G., Teklehaymanot, H. K., & Atsbha, G. M. (2024). Morphological Synthesizer for Ge’ez Language: Addressing Morphological Complexity and Resource Limitations.
• Teklehaymanot, H. K., Gidey, G., & Nejdl, W. (2025). Low-Resource English–Tigrinya MT: Leveraging Multilingual Models, Custom Tokenizers, and Clean Evaluation Benchmarks.

Publication list (PDF, 734 KB)

• Postgraduate Studies Coordinator of Aksum Institute of Technology (since 2024)
• Member of the Academic Council of Aksum Institute of Technology (since 2023)
• Secretariat of Department Committee, IT Department, Aksum University (since 2022)