Program leadership, conversion rate optimization, causal measurement, and growth analytics connected to decisions the business can defend.
Evidence in one sentenceAt NRG, that system scaled annual experiment velocity from roughly 20 to more than 100 tests; internal 2025 program reporting recorded $30M+ in impact.
Five regulated brands needed one way to turn ideas into comparable decisions. I built shared intake, prioritization, statistical rules, QA, readouts, and executive reporting so teams could move faster without quietly lowering the evidence bar.
Inspect the NRG programFor Startup Banking, I designed a comparative geo test, instrumented QR-enabled offline-to-online paths, and translated the result into an investment decision. The same role connected acquisition, product usage, CRM activity, and funded-account outcomes across a $1B+ pipeline.
Review the SVB measurement workGrowthLayer turns test plans, results, decision records, and reusable evidence into a searchable operating system. Building it forces the methodology to become product behavior: required identifiers, fail-closed decisions, analysis-only records, and reproducible planning.
Open GrowthLayerName what changes if the result is positive, negative, or inconclusive.
Size traffic, baseline, detectable effect, allocation, and decision horizon before creative work.
Define exposure, primary metric, guardrails, QA, and one prespecified decision method.
Record identifiers, limitations, business context, and the next decision—not only a winner.
AI increases delivery leverage. It does not replace the evidence standard.
I use coding agents to turn experiment policy into typed workflows, validation, tests, and decision records.
GrowthLayer decision workbench, calculators, and guarded ingestionModels help inspect patterns, draft readouts, and challenge assumptions; the metric definition and causal claim remain human decisions.
NRG analysis workflow and executive reporting systemAI workflows get test cases, structured rubrics, failure handling, and explicit boundaries before they earn autonomy.
Production publishing and product workflows with fail-closed gatesThe strongest fit is a role where experiment design, causal measurement, product judgment, and organizational adoption matter together. This page does not claim production ML-infrastructure or native SDK engineering depth.
Share the remit, decision rights, and outcome you need this role to own.
Discuss role alignment