AI Engineer – Multilingual LLM Development

About the role

AI Singapore (AISG) is seeking an AI Engineer to drive the development, training, evaluation and engineering of cutting‑edge multilingual, multicultural and multimodal foundation Large Language Models (LLMs). The role focuses on under‑represented languages and real‑world AI applications, working at the intersection of frontier research and productisation.

Key responsibilities

• Develop, train, fine‑tune and evaluate multilingual, multicultural and multimodal foundation LLMs, including pre‑training, supervised fine‑tuning, instruction tuning and model adaptation.
• Build and maintain scalable data pipelines for multilingual text ingestion, cleaning, filtering, deduplication, tokenisation and dataset mixture creation.
• Support large‑scale distributed model training workflows, covering experiment tracking, checkpointing, monitoring, debugging, profiling and compute optimisation.
• Review frontier AI and LLM research papers, reproduce methods and translate findings into practical improvements.
• Contribute reusable engineering assets such as training scripts, evaluation tools, model artefacts, APIs, documentation and deployment‑ready components.
• Collaborate with internal teams, external research labs, academic institutions and industry partners on LLM research and knowledge‑sharing activities.
• Participate in community‑building efforts through technical seminars, workshops and publications.

Required profile

• Relevant qualification (NITEC, Diploma, Bachelor’s, Master’s or PhD) in Computer Science, Artificial Intelligence, Machine Learning, Data Science, Engineering or equivalent practical experience.
• Strong interest in multilingual and multimodal AI research.
• Ability to work collaboratively across academia, industry and government partners.

Required skills

• Experience with large‑scale data pipelines for multilingual text.
• Proficiency in distributed model training and related optimisation techniques.
• Familiarity with experiment tracking, checkpointing and model evaluation workflows.
• Knowledge of LLM fine‑tuning, instruction tuning and preference optimisation.