Open source Data engineering
Modular ETL framework for African fintech
A reusable Airflow + dbt template for ingesting M-Pesa, Flutterwave, and POS data into a unified warehouse — with built-in quality tests and lineage.
AirflowdbtBigQueryPython
Selected work
Things we've shipped.
A selection of open-source projects, prototypes, and engagements that show how we work — across engineering, analytics, and AI.
Prototype Analytics
Real-time SME revenue dashboard
An open-source Streamlit dashboard SMEs can drop on top of their POS database — daily KPIs, customer cohorts, and a leakage detector.
StreamlitPandasDuckDB
Methodology Audit framework
Data-maturity audit framework
The two-week diagnostic we run with every new client — six dimensions, scored 0–4, with a prioritized 90-day roadmap as the output. Open-sourced as a template.
NotionSQLMethodology
Prototype Machine learning
Document summarizer for compliance teams
A local-context RAG system for summarizing long compliance documents, with citation, evaluation harness, and a guardrail layer for sensitive terms.
LangChainPineconeOpenAIPython
Open source Machine learning
Alternative-data credit scoring model
A gradient-boosted credit-scoring model that uses mobile-money transaction patterns instead of traditional credit history — with full SHAP explainability.
XGBoostSHAPscikit-learnFastAPI
Open source Machine learning
Demand-forecasting toolkit
A lightweight Python toolkit that benchmarks Prophet, ARIMA, and gradient-boosted regressors on retail demand data — with a reproducible CLI.
ProphetXGBoostPandas
More on GitHub
The full archive lives on GitHub.
Notebooks, reference architectures, and small-but-useful tools we've open-sourced over the years. Star, fork, or open an issue.
Like what you see?