Section 1 — Must‑have “Gate” Checks
Processing a résumé is blocked until hard‑stop gates are resolved. Use status codes to reflect progress and overrides.
Processing Status
OK
Hard‑stops: Right‑to‑Work, Critical Certification (if required), Identity/KYC
Gate | Status | Stop Rule | Notes |
---|
JD Gate Flags
If the JD requires a gate, failing it will prevent processing.
Section 2 — Scoring Buckets & Items
Create and tune multiple scoring profiles (Standard, AVAR, Client). Each section must total exactly 100%.
Parsing Fit (Content)
Total
0%
Item | Category | Percentage | What it checks |
---|
AVAR (Verification)
Total
0%
Item | Category | Percentage | What it checks |
---|
Modifiers & Evidence (± applied to Fit)
Total
0%
Item | Category | Percentage | What it checks |
---|
Tip: Modifiers do not need to sum to 100% universally, but each profile often targets ±10–20%. You can allow negatives for penalties.
Section 3 — Technical Reference: Scoring Logic
Mix α (Fit vs AVAR):
70% Fit / 30% AVAR
- Profiles & Weights. Fit items sum to 100%. AVAR items sum to 100%. Modifiers are ± applied to Fit (need not sum to 100%).
- Item signals. Fit item score ∈ [0–1] from résumé/JD evidence: presence × proficiency × recency × evidence. AVAR item: 1 (verified) / 0.5 (pending/expired) / 0 (failed).
- Gates → caps. Hard‑stops (Right‑to‑Work, Identity/KYC, required Cert) block on FAIL. Soft gates may cap Fit: PENDING ≈ 80%, MISMATCH ≈ 70%, EXEMPT = no cap.
- Fit. Base Fit = Σ(weight × item score)×100 → apply Modifiers → clamp [0,100] → Effective Fit = min(Fit+Modifiers, lowest active cap).
- AVAR. Weighted sum of verification outcomes ×100.
- Overall. Mix by policy: Overall = α·Effective Fit + (1−α)·AVAR.
// Rendering…