Case Study
Candidate Vetting and Interview Review Dashboard
Human decision-support dashboard for hiring. Combines job criteria, public profiles, resumes, portfolios, transcripts, assessments, and email history into a scored, explainable case with reviewer-only controls.
Decision-support map
Hiring evidence stays reviewable
Approved sources
FastAPI rubric
Scores + flags
Review dashboard
Human decision
System Summary
Built a hiring review dashboard that combines job requirements, OnlineJobs.ph and public profiles, resumes, portfolios, interview transcripts, assessments, approved email history, rate, schedule, and availability in one case.
Business Problem
Hiring reviewers had to cross-check job posts, profiles, resumes, interview notes, and messages manually, leading to inconsistent decisions and missed evidence.
What Czar Studios Built
FastAPI decision-support service, React/TypeScript reviewer UI, structured scoring across role, skill, communication, reliability, schedule, and pay-fit dimensions, plus strengths, risks, missing evidence, follow-up questions, and recommended next steps.
Architecture and Data Flow
Cases are scored server-side and presented in a compact card with an Analysis modal for breakdowns. Verdict and status enums preserve honest states such as insufficient evidence and needs evidence. Nonreviewer API access returns 403.
Integrations
OnlineJobs.ph, public profiles, resumes, portfolios, interview transcripts, assessments, and approved email/message history.
Permissions, Guardrails, and Human Review
Reviewer-only UI and API controls prevent autonomous contact, rejection, or hiring. Raw-secret rejection and decision-support-only boundaries are enforced. The system recommends, not decides.
Testing and Verification
Build and smoke verification passed 63/63 checks with live role/API readback.
Practical Outcome
Reviewers compare candidates using consistent evidence and explicit confidence, reducing manual cross-checking and making debatable calls easier to discuss.
Privacy-Safe Proof Notes
Dashboard and scoring structure only; no candidate names, contact data, or assessment content are exposed.