Elevate Your Data Strategy
At Devel AI, we build intelligent, data-driven systems grounded in validated machine learning research and real-world datasets. Our work spans analytics, visualisation, predictive modelling, and intelligent automation - all developed using your actual data behaviour, including its noise, patterns, and anomalies. This approach ensures every model we create is evidence-based, explainable, and capable of delivering measurable results.
Latest News
Research & Innovation
At Devel AI, our research bridges artificial intelligence, machine learning, and applied engineering to solve real-world problems, from predictive maintenance to biomedical intelligence and clean energy.
Machine Learning Approach to Investigate the Effects of Walking Speed on Gait Phase Sub-Durations
Authors: A.O. Bakare, A.A. Babashola, S.A. Animashaun, R.O. KehindePublished in: European Journal of Applied Science, Engineering & Technology (Vol 3, Issue 4, 2025)
Summary:
Developed a biomarker-driven gait-phase classification model using Random Forest and wearable sensor data.
The model achieved 100% classification accuracy, revealing how AI and biomechanics can support rehabilitation monitoring and early detection of neurological disorders.
🧩 Impact: Enables non-invasive gait analysis for tele-rehabilitation and clinical decision-support systems.
Predicting Proton Exchange Membrane Fuel Cell Performance through Advanced ML Techniques
Authors: A.O. Bakare, K.T. Seriki, S.M. Osunba
Published in: Scientia: Technology, Science and Society (Vol 2, No 8, 2025)
DOI: 10.59324/stss.2025.2(8).07
Summary:
This study applied ensemble learning (Stacking, Voting, and Bagging regressors) to predict real impedance in hydrogen fuel cells.
The Stacking Regressor achieved an R² of 0.9293 and RMSE of 0.1179, demonstrating how machine learning can optimise fuel-cell efficiency, reduce cost, and improve sustainability in hydrogen-energy systems.
🧩 Impact: Demonstrates how AI can accelerate green-energy innovation through predictive modelling.
Comparative Analysis of Gearbox Fault Detection Using Ensemble Learning Techniques
Authors: N.A. Raji, R.O. Kuku, A.O. Bakare, M.M. Ogunbiyi, T.I. Morafa
Published in: Journal of Production Engineering (Vol 27, No 2, 2024)
DOI: 10.24867/JPE-2024-02-001
Summary:
Introduced a robust machine-learning framework for industrial gearbox fault diagnosis using vibration-sensor data.
By comparing ensemble models (AdaBoost, Bagging, Stacking, Voting), the AdaBoostClassifier-ET achieved 87.56% accuracy, enhancing predictive maintenance and minimising downtime.
🧩 Impact: Supports smart-manufacturing systems and real-time machinery monitoring with low-cost sensors and ML.