F24PEOPLE

Resume screening with reasoned shortlists

For roles with high application volume, reads each resume against a structured definition of what the role actually requires, scores the fit, and produces a shortlist with reasoning for each ranking. Different from generic ATS scoring because the pattern works against an explicit role specification co-authored with the hiring manager, not against vague keyword matching, and because every shortlist comes with the why — the specific signals that put each candidate where they are. Pre-screens for disparate impact at the cohort level and refuses to make 'hire/no-hire' calls; final decisions stay with humans. The pattern's value is letting hiring teams give every applicant a thoughtful first look, even when volume would normally make that impossible.

WHERE THIS FITS

BUSINESS SHAPES

B2B servicesProfessional servicesProduct company

VOLUME THRESHOLD

Below 100 applications per role a month, the payback rarely earns the build. Patterns this shape reliably pay back at 500+.

FITS BEST

Firms hiring >5 roles/quarter, especially for non-cookie-cutter positions.

PAYBACK · 4-8 moBUILD · Low-MediumVALUE · $25k-$80kWHEN · >5 roles/quarter

FAILURE MODE TO DESIGN AROUND

Bias amplification. Training on past hiring reproduces past bias. Adverse-impact testing is gating, not optional.

REQUIREMENTS · 6 REQUIRED

Requirements describe capabilities the pattern needs in your environment, not the vendors you must buy. Any system that fills a requirement satisfies it — that’s what makes the catalog portable across the long tail of SMB tooling.

application_intake
REQUIREDREADevent
Where applications arrive with resume and any structured information from the application form.
DATA SHAPE
Per-application: resume document, cover letter if provided, structured form responses, application timestamp, role applied to.
COMMONLY FILLED BY
- applicant tracking system intake
- career site form submissions
- agency-sourced application portal
- referral submissions
role_specification_corpus
REQUIREDREADcorpus
Structured definitions of what each role requires. Critical: vague specifications produce vague scoring; specific specifications produce useful shortlists.
DATA SHAPE
Per-role: must-have qualifications, nice-to-have qualifications, disqualifiers, examples of strong and weak candidates, weighting between criteria.
COMMONLY FILLED BY
- role specifications co-authored during engagement with each hiring manager
- structured role library in the ATS
- internal role definition document refined per opening
shortlist_destination
REQUIREDWRITEbatch
Where the scored shortlist lands for the hiring manager and recruiter to act on.
DATA SHAPE
Ranked candidates with score, top reasons for the ranking, specific signals matched and missed, suggested interview focus areas.
COMMONLY FILLED BY
- shortlist view in the ATS
- weekly digest to the hiring manager
- dashboard with sortable candidates
bias_auditing_loop
REQUIREDREAD + WRITEbatch
Statistical monitoring of how the pattern's scoring distributes across protected characteristics (where lawfully captured), with flags for disparate impact.
DATA SHAPE
Aggregate statistics on score distributions vs. demographic data when available, monitored cohort-level rather than individual-level.
COMMONLY FILLED BY
- EEO data tracking in the ATS
- structured fairness review process built for the engagement
- monthly audit by people ops
human_decision_capture
REQUIREDREADevent
Final hiring decisions captured back so the pattern learns what actually predicts good hires vs. what predicts being shortlisted.
DATA SHAPE
Per-candidate: shortlist score, interview outcomes, offer status, ultimate hire status, optionally early-tenure performance.
COMMONLY FILLED BY
- ATS stage progression
- hiring decision records
- performance review data linked back over time
redaction_layer
REQUIREDREADrequest
Removes signals that should not influence scoring: names where used as proxies, photos, addresses to the extent they encode demographics. The redaction is for the model's view, not the recruiter's.
DATA SHAPE
Per-application processed view with sensitive fields blanked or hashed.
COMMONLY FILLED BY
- redaction module built for the engagement
- fairness-aware preprocessing in the ATS
- structured field stripping before scoring

RUNTIME FLOW · 8 STEPS

01
New application arrives through intake
application_intake
02
Apply redaction layer to strip signals that should not influence scoring
redaction_layer
03
Fetch the role specification for the role being applied to
role_specification_corpus
04
Score the application against the specification: must-haves, nice-to-haves, disqualifiers, with reasoning per criterion
05
Add to candidate ranking for the role
06
Periodically (per opening, or per batch threshold) publish the shortlist
shortlist_destination
07
Run bias auditing on the shortlist's distribution, flag if disparate impact is detected at cohort level
bias_auditing_loop
DECISION If disparate impact detected, pause or flag for human review.
08
Capture hiring decisions over time for learning what actually predicts good hires
human_decision_capture

EMISSIONS · 3

Structured outputs this pattern produces. Other patterns and client systems can subscribe to them, which is how the catalog composes over time.

shortlist_quality_signal
Per-role correlation between shortlist score and downstream outcomes (interview pass rate, offer extension, retention).
CONSUMED BY
- talent acquisition quality dashboards
- role specification refinement
- executive HR reviews
fairness_audit_signal
Cohort-level fairness metrics over time, the compliance evidence base.
CONSUMED BY
- People Ops fairness reviews
- legal compliance
- annual EEO reporting
role_specification_calibration
Where role specs seem misaligned with what actually predicts hire success.
CONSUMED BY
- talent acquisition team
- hiring manager retrospectives