Data pipelines and statistical analysis systems for international development.

What I Do

• Data Pipelines

  Config-driven, reproducible pipelines that ingest, transform, and harmonise survey and administrative data across sources. Built to extend, not to rebuild from scratch when the next round arrives.

  • Raw SPSS, CSV, and Stata files to analysis-ready datasets
  • JSON/YAML configuration per country-year, so the same codebase handles different surveys
  • Automated variable mapping, merging, and unique identifier creation
  • Used in: A&T equity analysis (15+ survey rounds, 3 countries), UNICEF JME screener (1,091 surveys, 167 countries)

• Code Migration

  Move legacy analysis from spreadsheets, SPSS, or ad-hoc scripts to production-quality Python or R. Preserve your statistical logic, improve everything around it.

  • Version control, testing, and documentation from day one
  • Reproducible outputs: same inputs, same results, every time
  • Integration with existing workflows (Stata control tables, team processes)
  • Used in: UNICEF JME screener (Python proof-of-concept migrating to R for Stata integration)

• Applied Statistics

  Methods your reviewers will trust, code your team can maintain. I handle complex survey designs and disaggregated analysis across demographic and geographic stratifiers.

  • Weighted survey analysis accounting for complex sampling designs
  • Logistic regression, PCA, mixed-effects models, Mahalanobis outlier detection
  • Equity decompositions by wealth, region, ethnicity, and education
  • Used in: A&T equity analysis (20+ years of MCHN trends disaggregated by 5 stratifiers), Namibia NHIES (multi-source consumption and nutrient analysis informing national fortification policy)

Case Studies