Three answers. One pipeline.
Fairness report, one click, per role and per quarter.
Per-role and per-stage adverse-impact analysis, 4/5ths rule applied per protected group, exportable as a PDF for your regulator. The report shows the rubric, the weights, the scores by group, the flags, and the redress actions taken. We publish our own audit summary annually on /trust/fairness.
↓ Decision engine
The rubric is explicit, defensible, calibrated.
The bar for a quant role, a risk role, a compliance role is encoded in a rubric framework you tune once and reuse across cycles. Senior people retiring does not lose the bar. The rubric is calibrated against eight years of psychometric science from Neuroworx, with construct-validity work documented in the model cards.
↓ Decision engine and assessment
EU data residency. Sub-processors short and inspectable.
Every service runs in London (Vercel lhr1, Supabase eu-west-2, LiveKit EU, Resend EU). The sub-processor list is on the trust hub with purpose, region, and DPA reference. We do not move candidate data to the US. SOC 2 Type II is in progress with Prescient Assurance and Drata.
↓ Infrastructure and trust hub