General Information

• Name: Mostafa M. ElAraby
• GitHub: github.com/mostafaelaraby
• Languages: Fluent in Arabic, English, and French (DELF B2)

Education

PhD in Computer Science (2020 - Present)

University of Montreal, Mila, Montreal, QC 

• PhD Candidate since Fall 2022.
• Research focused on out-of-distribution detection and generalization in autonomous driving.
• Received Excellence Scholarship from DIRO department at UdeM, Winter 2021.

M.Sc. in Computer Science (2019 - 2020)

University of Montreal, Mila, Montreal, QC 

• GPA 3.85 / 4.3.
• Thesis focused on neural networks compression using Mixed-Integer Programming.
• Received Excellence Scholarship from DIRO department at UdeM, Winter 2020.

B.Sc. in Computer Science (2009 - 2014)

Alexandria University, Alexandria, Egypt 

• Major in Computer Science with a Minor in Communications.
• GPA 3.3 / 4.0.

Experience

Research Intern (May 2023 - Dec 2023)

Precision AI (with Mitacs), Edmonton, AB 

• Researched out-of-distribution (OOD) detection within semantic segmentation, enhancing model reliability and safety.
• Developed a robust framework for assessing and improving the generalization capabilities of deep learning models.

Research Intern (Jun 2019 - Sep 2019)

Microsoft ATL, Cairo, Egypt 

• Led the development of a dialectical Arabic to English machine translation system.
• Worked on research related to adapting conversational context in neural machine translation.

Senior R&D Engineer (Sep 2018 - Mar 2019)

Tensorgraph 

• Researched and developed deep learning approaches for chatbot systems.
• Successfully improved and deployed the model to automotive dealers in Canada.

R&D Engineer (Nov 2016 – Jun 2018)

Microsoft ATL, Cairo, Egypt 

• Engineered a novel approach to generate dialectical data, significantly enhancing machine translation systems.
• Contributed to deploying an improved translation model to Skype Translator.

Software Engineer (Aug 2014 – Nov 2016)

Valeo, Cairo, Egypt 

• Developed MATLAB Simulink tools for automated model generation, reducing embedded C development time.

Publications

• GROOD: GRadient-aware Out-Of-Distribution detection, TMLR 2025 (ElAraby et al., 2025)
• Layerwise Early Stopping for Test Time Adaptation, AAAI 2025 (Sahoo et al., 2025)
• Background Aware Continual Semantic Segmentation, CRV 2024 (ElAraby et al., 2024)
• Identifying Critical Neurons in ANN Architectures, CPAIOR 2023 & NeurIPS OPT 2020 Workshop (ElAraby et al., 2020)

Notable Projects

• Continual Learning Challenge (CVPR 2021): Achieved 6th place on the leaderboard with team Real-DEEL.
• Imitation Learning for Lane Following (AI-DO 2019): Project selected as a baseline and merged into the duckietown gym.

Technical Skills

• Deep Learning: PyTorch, TensorFlow, Hugging Face
• Languages: Python, C++, Java, MATLAB
• ML Tools: scikit-learn, numpy, pandas, MLflow, Weights & Biases
• Infrastructure: Docker, Git, AWS, GCP, Hadoop, SQL, MongoDB

Hobbies

• Swimming
• Kayaking
• Travel

Supervisors

• PhD Supervisor: Prof. Liam Paull
• Masters' Supervisors: Prof. Margarida Carvalho and Prof. Guy Wolf

References

1. ElAraby, M., Sahoo, S., Pequignot, Y., Novello, P., & Paull, L. (2025). GROOD: GRadient-aware Out-Of-Distribution detection in interpolated manifolds. Transactions on Machine Learning Research (TMLR). https://openreview.net/forum?id=2V7itvvMVJ
2. Sahoo, S., ElAraby, M., Ngnawe, J., Pequignot, Y. B., Precioso, F., & Gagné, C. (2025). A layer selection approach to test time adaptation. Proceedings of the AAAI Conference on Artificial Intelligence, 39(19), 20237–20245.
3. ElAraby, M., Harakeh, A., & Paull, L. (2024). BACS: Background Aware Continual Semantic Segmentation. Proceedings of the Conference on Robots and Vision.
4. ElAraby, M., Wolf, G., & Carvalho, M. (2020). Identifying efficient sub-networks using mixed integer programming. OPT Workshop, 3.