Building AILEF: A Universal Engineering Framework for the AI Lifecycle

A developer at the École Supérieure Polytechnique of Dakar (ESP-UCAD) and UMMISCO has proposed SDLC-AI, a universal engineering framework for AI systems inspired by the Software Development Life Cycle. To operationalize it, they built AILEF (AI Lifecycle Engineering Framework), a platform that automates the entire AI lifecycle, addressing challenges in reproducibility, traceability, governance, deployment, monitoring, and continuous retraining.

Alhamdulillah I am pleased to share that I have successfully defended my Master's thesis in Artificial Intelligence and Big Data at the École Supérieure Polytechnique of Dakar ESP-UCAD . This research was conducted during my internship at UMMISCO International Joint Unit for Mathematical and Computer Modeling of Complex Systems . The Problem Today, building machine learning models has become easier than ever. Building reliable AI systems is a different challenge. While software engineering has relied for decades on standardized development processes such as the Software Development Life Cycle SDLC , AI projects are still often developed using ad hoc and empirical practices. As a result, many AI projects struggle with: • Reproducibility • Traceability • Governance • Deployment • Monitoring • Model maintenance • Continuous retraining Our Contribution To address these challenges, we proposed SDLC-AI, a universal engineering framework inspired by the traditional Software Development Life Cycle and adapted to the specific requirements of AI systems. To operationalize this framework, we designed and implemented AILEF AI Lifecycle Engineering Framework , a platform that supports and automates the entire AI lifecycle. The platform automates several lifecycle engineering tasks, allowing data scientists to focus on developing and improving machine learning models while the platform handles the underlying engineering complexity. What I Learned This project was much more than a Master's thesis. It gave me the opportunity to deepen my knowledge in: • Artificial Intelligence • Machine Learning • Deep Learning • Software Engineering • MLOps • ModelOps • DevOps • Explainable AI XAI • AI Governance • Model Monitoring • Continuous Retraining • Scientific research Acknowledgements I would like to express my sincere gratitude to Mr. Mandicou Ba, Dr. Fatou Ngom, Prof. Alassane Bah, Mr. Mouhamed Amar, the researchers at UMMISCO, the members of the examination committee, the faculty of ESP-UCAD, my family, my friends, and everyone who supported me throughout this journey. This achievement marks the end of one chapter and the beginning of another. I look forward to continuing my research and contributing to the development of more reliable, responsible, and industrialized AI systems. Thank you for reading.