General

• Embedding model: OpenAI's text-embedding-ada-002
• Vector Database: Postgres with pg_vector.

Preparation

• For each competency level in competency dictionary, get the embedding output and store in the database.

Inference Process

• For each item in "TUGAS, TANGGUNG JAWAB DAN HASIL AKHIR", get the embedding output and search the most similar competencies in the database.
• For each mandatory competency, get n-th most accurate results, where n = maximum total required competency.
• Adjust the competency level:
  • Visionary Leader: Lowest competency level = 3.
  • Strategic Leader: Lowest competency level = 3.
  • Operational Leader: Lowest competency level = 3.
  • People Leader: Lowest competency level = 3.
  • Team Leader: Lowest competency level = 3.
  • Principal: Lowest competency level = 3.
  • Senior Expert: Lowest competency level = 3.
  • Lead Specialist: Lowest competency level = 3.
  • Senior Generalist: Lowest competency level = 2
  • Expert: Lowest competency level = 2
  • Senior Specialist: Highest comptency level = 3.
  • Mid Generalist: Highest comptency level = 3.
  • Specialist: Highest comptency level = 2.
  • Junior Generalist: Highest comptency level = 2.