Applied Statistician — Data Vendor Validation & Test Design

Looking for a part-time applied statistician to help us design and run validation studies on third-party data vendors. The initial methodology is already written; we need someone sharp to run the analyses, pressure-test the math, and iterate. And build our confidence scoring for our data. We will have a lot of data to validate and analyze and we are looking to continuously improve in this regard so this could very will turn into consistent work beyond just a part-time thing. Required 3+ years applied frequentist statistics (validation studies, A/B tests, biostatistics, survey research, data quality assessment) Confident with binomial CIs (Wilson, Clopper-Pearson, normal approx — know when each applies) Stratified sampling + sample-size determination Python: pandas, scipy.stats. Not ML — no sklearn/pytorch needed. Read a methodology spec and immediately flag what's wrong with it Defensible interpretations of results, not vague hedging Our methodology: Four ground-truth groups (known-correct, known-wrong-person, known-dead-line, production-confirmed correct). Three nested rule ladders indexed on a confidence score. Class-conditional accuracy per band. False-rejection rates at any-confidence and high-confidence. Pass/fail thresholds with one-sided Wilson bounds. If you read that paragraph and instantly know what we're doing and what could go wrong, we want to talk.