Ndlelanhle Zondi

Built a scalable data pipeline architecture that extracts, cleans, and transforms multi-source commercial/operational transaction data in Python and PostgreSQL, processing over 1,000,000 messy rows per month with zero downtime. Translated high-level stakeholder questions into formal analytic problem statements with deterministic evaluation criteria. Architected and deployed 8 centralized executive Power BI dashboards with robust Star Schema models, reducing monthly data preparation and reporting cycle time by 45%. Developed automated data validation routines to assess completeness, uniqueness, and distribution stability (improving data quality scores from 85% to 99.2%). Performed exploratory data analysis to identify nonlinear relationships, distribution shifts, and seasonality across longitudinal records, and ran variance/“root-cause” analytics to detect KPIs, anomalies, and cost drivers. Optimized SQL using recursive CTEs, temporary tables, and window functions (e.g., RANK, LAG/L

Built a scalable data pipeline architecture extracting, cleaning, and transforming multi-source commercial and operational transaction data in Python and PostgreSQL, processing ~1,000,000 messy rows per month with zero downtime. Translated high-level stakeholder questions into formal analytic problem statements with deterministic evaluation criteria. Architected and deployed 8 centralized executive Power BI dashboards using robust Star schema models to reduce monthly data prep and reporting cycle time by ~45%. Developed automated data validation engines to assess completeness/uniqueness/distribution stability, improving data quality scores from ~85% to ~99.2%. Performed EDA with pandas/seaborn to identify non-linear relationships, shifts, and seasonality across longitudinal records; applied variance modeling to compare actual outcomes vs operational projections and identify performance anomalies and cost drivers. Optimized SQL queries using recursive CTEs, temporary tables, and window

Consolidated and standardized compliance datasets across 6 active enterprise infrastructure projects into unified analytical tables. Reconciled 450,000+ historical audit entries using advanced Excel, Power Query, and Python with deterministic rules to remove duplicate and invalid records. Implemented Python-based anomaly detection using Z-score and IQR statistical thresholds to flag abnormal spikes in equipment resource consumption and emissions logs. Produced recurring and ad-hoc executive scientific reports summarizing carbon emissions, energy usage, and safety compliance with precise uncertainty intervals. Conducted weekly requirements-gathering sessions with site engineers and environmental managers to refine sustainability data metrics and quantitative targets. Applied time-series decomposition to analyze multi-month resource trends and enabled procurement teams to optimize material purchasing schedules. Implemented strict data profiling and audit trails to ensure compliance with