As the Lead Investigator of AfriCrack: Benchmarking Domain-Shift Robustness of Crack Segmentation Models in African Infrastructure study. I managed the full research workflow, including literature review, project scoping, dataset creation, and scientific writing. I led a team that collected and annotated 1,000 crack images across Nigeria using Roboflow, and trained CNNs alongside the SegFormer Vision Transformer using PyTorch. The evaluation revealed a ~50% performance drop for models trained on non-African datasets such as SDNET2018, underscoring the importance of context-specific data. The project received an honorary mention at Deep Learning Indaba 2024 and placed 1st runner-up at IndabaX Nigeria 2024.

• View Manuscript

I carried out a computational study titled on the Determination of the Natural Frequency and Mode Shape of Beams Using Analytical and Finite Element Methods with Python. I implemented beam deflection functions using the Rayleigh–Ritz method and the Principle of Virtual Work, and developed Python routines to compute natural frequencies for multiple end conditions, including clamp–clamp and simply supported configurations. This work strengthened my experience in structural dynamics, numerical modeling, and applied computational methods for engineering analysis.

• View Project