GENERATIVE AI-BASED SMART MANPOWER PLANNING AND CAREER DEVELOPMENT SYSTEM FOR CIVIL SERVANTS

Abstract

This paper presents a manpower planning and career development system for civil servants in a Thai government agency, where promotion eligibility was previously checked by hand from paper records. The system encodes the agency's career path and promotion criteria as machine-readable conditions. It was built with Next.js and TypeScript over a PostgreSQL database hosted on Supabase, which holds 18 tables, 16 of them encoding the agency's promotion criteria, and it runs in two passes. The first pass is a rule engine that builds Dynamic SQL queries, checks each officer against 13 target positions, scores readiness from academic qualifications, tenure, KPI assessment results and completed training, and returns a gap report as JSON. The second pass sends that report to a Gemini Flash Lite language model, which writes it up as a Thai individual development plan. The model cannot alter any score or eligibility decision, and every output stays in draft until human resource staff approve it. A System Usability Scale survey of 20 users returned a mean score of 78.5, which falls in the acceptable range.

Keywords: Manpower Planning, Career Path, Generative AI, Rule Engine, Individual Development Plan