Two years on the data behind 1,200+ telecom mast sites: a generator fault model, automated reporting, and the KPI definitions every team reported against.

Makasa Sun provides solar and hybrid power for telecom mast sites. I spent two years on the data behind their sites: what was failing, what it cost, and how leadership saw it.

Weekly reporting for 1,200+ sites was assembled by hand in Excel. Engineer visits were driven by faults after they happened.

• A fault model on fault-code history. A generator fault model (ARIMA and XGBoost) flagged sites due for an engineer check-up from fault-code history at 87% precision.
• Pipelines instead of spreadsheets. Weekly reporting for 1,200+ telecom mast sites moved from manual Excel to Python scripts and automated pipelines (SQL, AWS S3, Redshift), cutting weekly reporting time from 40 hours to 10.
• Clean data before clever models. Maintained data quality with SQL validation checks that caught bad readings before they reached reports, and investigated sites that stopped sending data.
• One set of numbers. Set the standard KPI definitions every team reported against and built the Power BI and Tableau dashboards leadership used for capital allocation.
• Payments. Analyzed 50K+ monthly site payment transactions in Python, improving forecast accuracy by 10% and reducing revenue variance by $120K a year.
• Ownership. Led the predictive maintenance project end to end, agreeing deliverables and timing with leadership, and managed a 3-person finance team.

• The results were used in planning a 200+ site expansion worth $2.1M in projected revenue.
• Weekly reporting: 40 hours → 10. Weekly reporting for 1,200+ telecom mast sites moved from manual Excel to Python scripts and automated pipelines (SQL, AWS S3, Redshift), cutting weekly reporting time from 40 hours to 10.

Makasa Sun. Diagrams on this page are mine and contain no company data.