Ogochukwu Okafor's Portfolio

Ogochukwu Okafor

Data Scientist | Machine Learning Engineer | AI Trainer

I don’t just analyze data I uncover the stories behind it.

Professional Summary

    • I have always liked working with data and finding out what the numbers are really saying. My background in Economics first got me interested in patterns and decision making, and that later led me to study Data Science and Artificial Intelligence at master’s level. Since then, I have worked on projects across data analysis, machine learning, business intelligence and AI, using tools like Python, SQL, Excel, Power BI, Power Query, DAX, Azure, Databricks and Docker.

    I also have hands-on experience with Generative AI, large language models, prompt design, AI evaluation, synthetic data, debugging and unit testing. I enjoy solving problems, learning new tools and turning data into something useful. At this stage, I am looking for opportunities where I can keep building my skills in data science and AI while working on practical projects that have a clear purpose.

    Key Skills

    Python Power BI Excel SQL Microsoft Azure Generative AI Large Language Models (LLMs) Tableau

    Business Insight 360

    Designed a multi-view Power BI dashboard for AtliQ Hardware, analyzing finance, sales, marketing, supply chain, and management, by transforming 1.8M+ MySQL and Excel records to improve efficiency by 30% with DAX Studio and support data-driven decisions.

    Bitcoin Future Price Prediction

    This project employs machine learning techniques, primarily utilizing the historical Bitcoin-GBP data spanning from January 1st, 2015, to September 1st, 2023.Four models – Multilayer Perceptron (MLP), Long Short-Term Memory (LSTM) network, Random Forest Regressor, and Gradient Boosting Machine Regressor were developed to predict Bitcoin prices. The top two performers, MLP and Random Forest, from both artificial neural networks and ensemble models were selected and optimized for future price predictions, yielding promising results exceeding 99% accuracy.

    Semantic Segmentation: Exploring FCN, PSPNet, and UNET Architectures

    This project delves into the comparative efficacy of three prominent architectures: FCN, PSPNet, and U-Net. By implementing these models and analyzing their performance, I gain profound insights into their strengths and limitations. Exploring augmentation techniques, I uncovered a plethora of methods for manipulating and presenting images, enhancing their diversity, and enriching our understanding of segmentation challenges.