Public Sector / Government Recruitment

Case Study: AI-Powered Job Description Generator for Bilingual Recruitment

AI Infrastructure & NLP Deployment

@2023

120% Increase in revenue ROI

AI Infrastructure & NLP Deployment

@2023

120% Increase in revenue ROI

Overview

A government agency responsible for high-volume recruitment sought to improve its job description authoring process. Traditionally, job roles were documented manually in Arabic and then translated into English — a labor-intensive process prone to inconsistency and delay. Additionally, sensitive HR data could not be processed via public cloud solutions due to national data protection laws.

The client’s core challenge was to accelerate and standardize bilingual job description generation, while ensuring full data sovereignty and operational control.

Objectives

  • Automate the generation of Arabic and English job descriptions

  • Comply with national data privacy and localization mandates

  • Reduce manual effort for HR teams and improve consistency

  • Operate entirely on local, secure infrastructure

  • Provide customization options for different departments and job types

Solution Delivered

Iterative Research Tech deployed a lightweight, CPU-optimized Job Description Generator, powered by a custom-trained local large language model (LLM) hosted on the agency’s internal infrastructure.

The solution accepts high-level job role inputs (in Arabic) and generates full job descriptions in both Arabic and English. It includes an admin interface for content customization, classification by role type, and template management — all fully localized.

Technical Highlights:

  • Multilingual Natural Language Understanding:
    Fine-tuned for domain-specific HR language in both Arabic and English. Capable of accurately mapping vague role inputs to formal job description outputs.

  • Local LLM Deployment (CPU-Based):
    Engineered for full inference capability on CPU-only hardware. Removed need for GPU or external compute dependencies.

  • No Internet / Cloud Dependency:
    Operates fully offline, ensuring compliance with government-mandated data sovereignty requirements.

  • Custom Template Engine:
    Departments could select pre-built templates for tone, length, and regulatory structure (e.g. probation clauses, grading bands, etc.).

Results

MetricBefore ImplementationAfter ImplementationAverage Time per JD90–120 minutesUnder 10 minutesHR Team InvolvementFully manual70% automation achievedLanguage Accuracy (QA)78%98.2% human-rated accuracyDual-language ConsistencyVariableFully synchronized outputInfrastructure Cost (Cloud)High / Blocked by policyZero additional cost (CPU-based)

Impact

  • Operational Efficiency:
    Cut average time to produce bilingual job descriptions by over 85%, allowing HR teams to support faster hiring cycles.

  • Regulatory Compliance:
    Met stringent local data policies by avoiding cloud solutions entirely. Internal IT and legal teams cleared the tool for use across agencies.

  • Scalability:
    Enabled HR to support new digital hiring campaigns without expanding staff — yielding a 120% increase in ROI on recruitment processing.

  • Localization Leadership:
    Positioned the agency as a regional first-mover in Arabic-first AI solutions for public service.