I’m Reginald Erzoah, a Data Professional building data systems, open-source tools, and developer-focused solutions for modern data ecosystems.
I work across machine learning, analytics, data tooling, and data workflows, with a focus on practical systems that help teams better understand, validate, and work with data.
I’m actively building and maintaining open-source projects like Dift and Aniwa, focused on solving real-world data quality, profiling, and dataset comparison challenges for data professionals.
My work combines machine learning engineering, data science, and product-minded thinking to create tools that are technically reliable, accessible, and useful in real-world environments.
I’m especially interested in ML infrastructure, MLOps, intelligent data systems, and open-source collaboration that improves how data and ML systems are built and maintained.
Featured Data Projects
[ML SYSTEM]
Credit scoring system that supports risk prediction, model interpretability, and interactive loan decision analysis through a production-style Streamlit interface.
[ML SYSTEM]
An interactive customer intelligence system that applies clustering workflows and behavioral segmentation to help teams better understand customer patterns and business value distribution.
[ML SYSTEM]
A fraud detection system that enables interactive transaction analysis, configurable prediction thresholds, and explainable machine learning inference for suspicious activity detection.
[BUSINESS ANALYSIS]
An end-to-end Business Analysis case study for a mobile-first recurring bill payment platform, demonstrating business analysis, requirements engineering, user research, UX design, process modelling, and Agile delivery.
Featured Side Quests
Open-source Python CLI tool for comparing datasets and identifying schema, row-level, and data quality differences.
View Repo
Universal data profiling and intelligence tool for inspecting datasets, diagnosing quality issues, and supporting modern data workflows.
View Repo
Fast-paced offline arcade game where you survive a stream of corrupted data while working as a glowing Data Node.
Available as a standalone python game and also a vscode extension.
View RepoFeatured Insights
Lessons from designing Python command-line tools that stay modular, testable, and maintainable as they scale.
Moving from experimentation into real-world machine learning systems, workflows, and production-ready tooling.
A practical look at profiling, validation, and trustworthy datasets before model training begins.
