AI Automation and NLP Engineering Intern

The Environment and Development Division seeks an intern to support an internal automation platform that converts UN meeting media into structured, verified reports using local large-language-model inference. The intern will contribute to a Python codebase focusing on software quality assurance, model evaluation, and tooling development, with a fully remote or Bangkok-based work arrangement possible.

Responsibilities

• Author and extend automated tests (unit and mocked-LLM tests), lifting coverage across priority modules and hardening pure functions and data-extraction paths.
• Design and run structured model and prompt evaluations, such as per-language machine-translation selection and speech-recognition error detection, including building labelled gold sets for measurable results.
• Build and extend deterministic extraction tooling (pattern and catalog matching, fuzzy-match resolvers) consistent with the project's principle that language models never supply structured identifiers.
• Scaffold new modules and utilities to the house architecture, and improve documentation, refactoring and developer tooling.

Requirements

• Be enrolled in, or have completed, a graduate school programme (second university degree or equivalent, or higher) related to computer science, data science, artificial intelligence, machine learning, computational linguistics, software engineering, or related disciplines; or be enrolled in, or have completed, the final academic year of a first university degree programme (minimum bachelor's level or equivalent) in related fields.
• Strong Python proficiency, including experience with testing frameworks (pytest) and version control (git), is required.
• Foundational understanding of natural-language processing and of large-language-model prompting and evaluation is required.
• Familiarity with running or calling local LLMs (e.g. Ollama, Hugging Face) is an asset.
• Experience designing evaluation harnesses or working with structured (JSON) model outputs is an asset.
• Clean, documented and reproducible coding practice is essential, as contributions are made within an existing codebase and standard.
• Familiarity with the responsible use of AI tools to improve work quality and efficiency, in line with UN guidelines, is desirable.
• Fluency in spoken and written English is required for the Internship Programme.
• Knowledge of another language used in Asia-Pacific is an asset (e.g. Chinese, Vietnamese, Bahasa Indonesia, Hindi, Urdu, Thai).

Skills

• Python Programming
• PyTest
• Git Versioning
• Natural-language processing
• Large-language-model prompting
• Model evaluation
• JSON data handling
• Software quality assurance
• Clean coding practices
• Reproducible coding
• Responsible AI Principles
• Computational linguistics
• Software Engineering
• Automation platform development

Languages

English, French