LLM Data Engineer for Foundation Models

Institute of Foundation Models; Abu Dhabi; On-site; AED 240,000 - 420,000; Full time

Job summary

The Institute of Foundation Models in Abu Dhabi invites an AI Engineer focusing on LLM data to join a world-class research team. You will build and refine training data for foundation models across pre-training, mid-training, and post-training phases, contributing to robust datasets and efficient pipelines.

You will collaborate with researchers and engineers, implement data quality controls, and contribute to reproducible results, documenting methodologies and findings for knowledge sharing.

Qualifications

• Proficient Python development for research workflows.
• Hands-on ML/NLP/LLM experience.
• Experience with large-scale data processing and QC.

Responsibilities

• Build and curate large-scale datasets for LLM training (pre/mid/post).
• Develop and optimize data pipelines: extraction, cleaning, filtering, deduplication.
• Create LLM-based data synthesis and augmentation workflows with prompt-based generation.
• Research methods to improve data quality, diversity, coverage, and efficiency.
• Design experiments linking data and model performance to guide improvements.
• Build scalable tools for processing and analyzing big datasets.
• Analyze data for quality issues, biases, duplication, and distribution gaps.
• Collaborate with researchers and engineers to translate needs into data solutions.
• Document datasets, experiments, and findings to ensure reproducibility.

Skills

Python

ML/NLP/LLMs

Data analysis

Collaboration

Tools

PyTorch

Hugging Face

vLLM

Job description

The Institute of Foundation Models in Abu Dhabi invites an AI Engineer focusing on LLM data to join a world-class research team. You will build and refine training data for foundation models across pre-training, mid-training, and post-training phases, contributing to robust datasets and efficient pipelines.

You will collaborate with researchers and engineers, implement data quality controls, and contribute to reproducible results, documenting methodologies and findings for knowledge sharing.