gabogg/semillero-2-AI-ATS
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
semillero-2-AI-ATS/docs/discovery-and-scope.md

3 KiB
Raw Blame History

Phase 1: Discovery & Scope

This document establishes the business boundaries, user personas, backlog of user stories, and Definition of Done (DoD) for the AI Recruitment Platform.

User Personas

  • Technical Recruiter: Focuses on sourcing, intake, initial AI-assisted screening, seniority verification, and tracking candidates through the hiring pipeline.
  • Hiring Manager: Focuses on reviewing pre-screened and scored candidates, evaluating role suitability, and conducting structured interviews.
  • Candidate: Interacts with the platform to submit applications/CVs and receives automated status updates.

Prioritized Backlog of User Stories

  1. CV Intake
    • User Story: As a recruiter, I want to upload candidate PDFs via the Web UI so they can be processed and stored.
    • Acceptance Criteria:
      • Drag-and-drop or file selector for PDF CVs.
      • Successful text extraction and ingestion.
      • Parsing and storage of extracted candidate info.
  2. Semantic Ranking
    • User Story: As a recruiter, I want a ranked list of candidates matching an open job description based on contextual relevance.
    • Acceptance Criteria:
      • Compute cosine distance between job embeddings and candidate embeddings.
      • Display ranked matches with normalized similarity score between -1 and 1.
  3. Profile Summary
    • User Story: As a recruiter, I want a terse AI-generated candidate summary to speed up screening.
    • Acceptance Criteria:
      • Display a short summary on candidate profiles.
      • Synthesized from parsed resume details.
  4. Seniority Guard
    • User Story: As a recruiter, I want automatic seniority detection to route candidates to the correct interview pool.
    • Acceptance Criteria:
      • AI-based classification (e.g., Junior, Mid, Senior).
      • Correct pool routing matches candidate seniority level.
  5. Unified Scoring
    • User Story: As a hiring manager, I want to compare candidate suitability using a standardized score.
    • Acceptance Criteria:
      • Standardized scoring schema with AI evaluation.
      • JSON structure containing keys: summary, classification, suggestions, and riskLevel.
  6. Stage Progression
    • User Story: As a recruiter, I want candidate stages to update dynamically, triggering confirmation emails.
    • Acceptance Criteria:
      • Visual stage progression pipeline.
      • Updating stage triggers background event notifications.

Definition of Done (DoD)

  • Type Safety: Fully typed Next.js App Router with TypeScript (no any types where avoidable).
  • Database Schema: Supabase relational database schema complete with verified pgvector distance metrics.
  • Linting & Quality: Zero TypeScript compilation or linting warnings/errors.
  • Git Quality Gates: Active Git quality gates (hooks) verifying builds, conventional commit messages, linting, and formatting.