Skip to content

Modeling

Table of contents

Product objective

Models support identification and explanation of retention risk while preserving human review.

Prediction boundary

Every model must define:

text
population
prediction target
time horizon
feature cutoff
exclusion criteria
decision context

Evaluation

Evaluation should include:

text
discrimination
calibration
precision and recall
threshold behavior
temporal validation
subgroup analysis
business interpretation

No single metric should define readiness.

Explainability

Outputs should include structured evidence:

text
risk score
important signals
confidence or uncertainty
data limitations
recommended next review

Resume Intelligence

Resume Intelligence may provide structured context extraction and transparent signals. It must not infer unsupported personality or psychometric traits.

Retention Advisor

The advisor combines validated signals, evidence, policy, and optional provider output into reviewable recommendations.

The advisor is decision support, not an autonomous employment decision-maker.

Limitations

Psychometric Intelligence is outside v1.0 until supported by appropriate data, validated instruments, fairness analysis, privacy review, and responsible interpretation.

Built with VitePress and deployed with GitHub Pages.