Talent Systems — Science Team
Your Immediate Workflow

What to Do First

Your most impactful actions right now, in priority order.

Priority 1: Review Existing Defaults (BLOCKS onboarding)

Time estimate: ~1 week

The platform has seeded defaults for every configurable field. These defaults were written by engineering based on I-O psychology principles, but they have NOT been reviewed by a scientist. Before any real employer uses the platform, you must:

Scorer Rubric (most critical)

Go to staging.talentsystems.ai/admin/rubrics and review:

  1. Scoring philosophy — "Score conservatively. A 3 is solid..." Is this the right anchoring?
  2. Ownership rule — "Exclusively collective language scores no higher than 2..." Too strict? Too lenient?
  3. Specificity rule — "Vague answers without a specific situation..." Appropriate threshold?
  4. Quantification bonus — "Quantified outcomes tip borderline 3-vs-4 toward 4..." Valid?
  5. Coached response flag — "Suspiciously polished answers..." What defines coached?
  6. Adverse impact guidance — "If any question touched a protected characteristic..." Correct trigger?
  7. Advance threshold — Currently 3.5. Should it be higher/lower?
  8. Hold threshold — Currently 2.5. Should it be higher/lower?
  9. Scoring rules — The add/remove rule list. Any to add? Any to remove?

Interviewer Rubric

Go to staging.talentsystems.ai/admin/interviewer-prompt and review:

  1. System identity — How the AI introduces its role
  2. Tone — "Professional but conversational" — appropriate?
  3. Probing style — "Curious and direct" — appropriate?
  4. Pacing — Candidate-finishing rules
  5. Opening/closing scripts — How the interview starts and ends
  6. Hard rules — The guardrails list
  7. Max probes per question — Currently 1. Enough? Too many?

Competency Anchors

Go to staging.talentsystems.ai/admin/competencies and review all 8:

  • Problem Solving, Ownership, Communication Clarity, Collaboration
  • Learning Agility, Resilience, Analytical Thinking, Initiative

For each: Are the 5-level anchors correctly calibrated? Does level 3 represent "meets expectations"? Does level 5 require truly exceptional evidence?

Integrity Weights

Go to staging.talentsystems.ai/admin/integrity and review:

  • Deduction weights per flag type (multiple faces, gaze away, tab hidden, etc.)
  • Gaze angle and duration thresholds
  • Human review threshold

Priority 2: Adaptive Rubric Design (BLOCKS adaptive mode)

Time estimate: ~1 week

The full questionnaire is at Phase 9 Questions. The most critical deliverable is the complete adaptive rubric JSON (Section 5 of that document). This JSON seeds the database and becomes the default configuration.

Priority 3: Validity Schema Review (BLOCKS data collection)

Time estimate: ~3 hours

Four specific questions at Phase 13 Questions. Quick turnaround, high impact — every interview from this point forward will contribute to the evidence base.

Priority 4: Posting-Level Overrides

Time estimate: ~2 days

Which rubric fields should employers be able to override per posting? Which admin fields should employers NOT see? This is a Dr. Rigos decision item.

Dependency Map

Which features depend on which science deliverables, and in what order.

Ongoing Responsibilities

What the science team owns on an ongoing basis after initial setup.

On this page

Priority 1: Review Existing Defaults (BLOCKS onboarding)Scorer Rubric (most critical)Interviewer RubricCompetency AnchorsIntegrity WeightsPriority 2: Adaptive Rubric Design (BLOCKS adaptive mode)Priority 3: Validity Schema Review (BLOCKS data collection)Priority 4: Posting-Level Overrides