Rancang Bangun Aplikasi Web Penjadwalan Mata Pelajaran Menggunakan Genetic Algorithm Dan Tabu Search Dengan Fitur Chatbot Berbasis Large Language Model (LLM) (Studi Kasus: Smk Bim Jombang)

Course timetabling at SMK Bakti Indonesia Medika (BIM) Jombang has traditionally been conducted manually using Microsoft Excel, requiring two to three days and remaining vulnerable to human error that causes schedule conflicts. This study aims to design and build an automated web-based course timetabling system using React, Golang, and PostgreSQL by implementing a Hybrid Genetic Algorithm (GA) and Tabu Search (TS) approach integrated with a Large Language Model (LLM) chatbot via Telegram. The hybrid GA-TS approach optimizes schedule solutions to satisfy all mandatory rules (hard constraints) and minimize preference violations (soft constraints), while the Gemini API-powered chatbot utilizes in-context learning to interactively present active schedules in natural language. Functional evaluation using Blackbox Testing across 34 scenarios shows that 100% of system features operate validly according to specifications. Furthermore, algorithm performance testing across 10 trials indicates that the Hybrid GA-TS model achieves an optimal fitness score of 1.0000 in 9 out of 10 trials (average fitness of 0.9143) with an average convergence time of 181.17 seconds, successfully producing conflict-free schedules.