RoleNorthSign in
System overview

How RoleNorth turns a résumé into a career compass.

A structured pipeline — parse, reason, score, recommend, store — built so every report is reproducible, auditable, and comparable month over month.

End-to-end flow

Upload

PDF / DOCX

Extract

Text + structure

Reason

Claude prompt

Score

4 dimensions

Recommend

3 pivots

Persist

User history

Deliver

Dashboard + PDF

Scoring engine

Four dimensions calibrated against industry datasets.

  • Skill decay

    decay = Σ(skill_age × industry_velocity) / total_skills

  • Automation exposure

    exposure = weighted(task_automatability, role_centrality)

  • Pivot readiness

    readiness = transferable_skills ∩ target_roles / target_requirements

  • Market demand

    demand = log(open_roles) × salary_premium × geo_match

Rules engine

Deterministic gates running before and after the AI step.

  • IF tier ∈ {basic, pro, transition} AND previous_report exists
    THEN include comparison_block
  • IF automation_exposure ≥ 70 AND employment = employed
    THEN upweight pivot recommendations
  • IF tier = transition
    THEN generate 90-day roadmap + 1:1 booking link
  • IF resume_text < 500 chars
    THEN reject with parse_error → user retry

Data model

users

  • · id
  • · email
  • · auth_provider
  • · plan
  • · created_at

reports

  • · id
  • · user_id
  • · tier
  • · scores
  • · pivots
  • · action_plan
  • · created_at

comparisons

  • · id
  • · user_id
  • · report_a
  • · report_b
  • · delta_scores

subscriptions

  • · id
  • · user_id
  • · stripe_id
  • · tier
  • · renewal_at

resume_blobs

  • · id
  • · user_id
  • · filename
  • · storage_path
  • · parsed_text

audit_log

  • · id
  • · actor
  • · action
  • · target
  • · ts

Security

Row-level isolation. Résumé blobs encrypted at rest. Social-only auth.

Reproducibility

Every report stores prompt version, model, seed, and inputs.

Latency

Median analysis < 12s. Background queue absorbs spikes.