Professional Experience & Projects

A comprehensive log of my internships, open-source contributions, and research initiatives.

Featured Academic Research

B.Sc. Final Year Project: Multimodal Breast Cancer Detection

@ Landmark University (Dept. of Computer Science)

Research Scope: Addressed the limitations of single-modality screening by designing a system that combines Ultrasound Imaging with Clinical & Radiometric data.

Methodology & Tech Stack:
  • MedCLIP (Vision): Employed a vision-language model to extract 512-dimensional semantic embeddings from ultrasound images.
  • TabNet (Tabular): Utilized an attention-based network to process clinical metadata and radiometric features.
  • Early Fusion Strategy: Concatenated image embeddings with 64-dimensional tabular embeddings to create a unified vector.
  • Performance: Achieved 82.7% Accuracy and 87.1% ROC-AUC, significantly outperforming unimodal baselines.

Professional Timeline

Machine Learning Intern

@ Kontemporary Konsulting Ltd.

My Role: Focused on researching and implementing deep learning architectures for medical diagnostics.

Key Deliverables:
  • Brain Tumor Classification: Benchmarked Vision Transformers (ViT) against DINOv2 architectures. Implemented Grad-CAM visualization.
  • Kidney Disease Prediction: Built and deployed a Random Forest classifier with full preprocessing pipelines.

IT & Automation Intern

@ TETFund

My Role: Tasked with identifying workflow bottlenecks and modernizing legacy data processes.

Key Deliverables:
  • Workflow Automation: Wrote Python scripts to automatically parse, clean, and classify institutional reports.

Featured Code Repositories

Global Tech Salary Prediction

Python • Regression • Pandas

A regression analysis model predicting tech salaries based on global job metadata and experience levels.

Multimodal Cancer Detection

PyTorch • MedCLIP • TabNet

Codebase for my final year project implementing the Early Fusion strategy for breast cancer detection.

Human Activity Recognition

Time-Series Transformers

End-to-end activity recognition system processing wearable sensor data on the HAR-70 dataset.

Heart Disease Prediction

Scikit-Learn • RandomForestClassifier

Classification model featuring extensive data cleaning, feature selection, and ROC/AUC evaluation metrics.

Bunmi Adelugba & Co.

HTML/CSS • Client Project

Official website for a Chartered Accounting firm. Features animations, responsive layout, and brand integration.

C# Calculator

Frontend Logic

A functional, responsive digital calculator demonstrating core Net.Framework and UI logic.