// computer vision · explainable ai
Data Scientist & ML Researcher
I build machine learning systems that don't just perform well — they explain themselves. My research probes what Vision Transformers actually see; my applied work turns that same rigor into churn models, fraud detectors, and recommendation engines that ship.
model output, not a filter — see Research
// about
I'm a computer science graduate from Benin working at the intersection of applied machine learning and explainability research.
My path runs through applied ML research at Covenant University in Nigeria and real-world data projects that bridge theory with impact. That combination shows up in everything I build: research questions grounded in production concerns, and production systems held to research-grade standards of evaluation.
My published work spans explainability methods for Vision Transformers and bibliometric analysis of machine learning benchmarks — two ends of the same question: can we trust what a model tells us, and can we trust what the field says about itself?
Outside research, I build end-to-end pipelines that work: predicting customer churn, detecting fraud, recommending content — always aiming for systems that are accurate, interpretable, and actionable.
// skills
// projects
End-to-end ML application predicting customer churn in the telecom sector. Covers preprocessing, feature engineering, and classification models (logistic regression, random forest), evaluated with ROC-AUC on imbalanced data and served through a real-time Streamlit dashboard.
View on GitHubML system detecting fraudulent telecom transactions from behavioral and call-pattern features. A modular pipeline from data ingestion to deployment, with a Streamlit interface for real-time fraud prediction and decision support.
View on GitHubContent-based recommendation engine using cosine similarity over vectorized text features. Applies vector space models and information-retrieval fundamentals through an interactive Streamlit interface.
View on GitHub// research
International Conference on Image Information Processing (ICIIP) — Jaypee University
Investigates the robustness and fidelity of explainability methods for Vision Transformers under image degradation, using perturbation-based evaluation metrics including insertion, deletion, and max-sensitivity.
Read the paperNIPES — Journal of Science and Technology Research
A large-scale bibliometric and trend analysis of benchmark datasets in machine learning research, examining usage patterns, citation impact, and the field's evolution through statistical and data-driven methods.
Read the paper// experience & education
Intensive training program in data science and machine learning.
Covenant University, Nigeria · GPA 4.38 / 5
University of Abomey-Calavi (IMSP) · GPA 3.5 / 5
// contact
I'm open to data science roles, research collaborations, and freelance ML projects. If you have a problem worth solving with data, let's talk.
🇧🇯 Abomey-Calavi, Benin — open to remote, worldwide