Pretraining Data

Overview

Microsoft AI is looking for Researchers and Research Engineers to join our Pre-training Data team and help train the world’s most capable AI frontier models, pushing the boundaries of scale, performance and product deployment. The Pre-Training team at Microsoft AI tackles some of the most challenging problems in deep learning at scale. As a team, we will deliver one of the best foundation models in the world, forming the foundation of many initiatives across Microsoft AI.
In this role, you’ll work at the intersection of data and innovation, collaborating with scientists, engineers, and annotators to curate, analyze, and evaluate diverse and large-scale data sources critical to model development. You will lead efforts to:

• Design and curate high-quality datasets to power frontier AI models.
• Develop novel data collection strategies.
• Improve dataset quality and integrity.
• Understand data-driven model behaviors.
• Align datasets with ethical and societal values.

This is a cross-disciplinary, high-impact role ideal for engineers who want to push the boundaries of what AI can learn from data. Our focus is on research and engineering that improves the models and products we ship. Academic publishing is not a priority for this role.

Responsibilities

• Create high-quality datasets for training and evaluation; run experiments on new datasets (data ablations) to assess their impact and determine the most effective data.
• Develop and maintain scalable data pipelines for ingestion, processing, filtering, and annotation.
• Analyze real-world datasets to assess quality, diversity, relevance, and identify areas for improvement.
• Build lightweight tools and workflows for dataset auditing, visualization, and versioning.
• Collaborate with Safety, Ethics, and Governance teams to ensure datasets meet standards for quality, privacy, and responsible AI practices. 

Qualifications

Required Qualifications:

• Bachelor's Degree in AI, Computer Science, Data Science, Statistics, Physics, Engineering, or related technical discipline AND 4+ years technical engineering experience with coding in languages including, but not limited to, Python and common data libraries (Pandas, NumPy, etc.)
• OR equivalent experience.

Preferred Qualifications:

• Bachelor's Degree in AI, Computer Science, Data Science, Statistics, Physics, Engineering, or related technical discipline AND 10+ years technical engineering experience with coding in languages including, but not limited to, Python and common data libraries (Pandas, NumPy, etc.)
• OR Master's Degree in AI, Computer Science, Data Science, Statistics, Physics, Engineering, or related technical discipline AND 8+ years technical engineering experience with coding in languages including, but not limited to, Python and common data libraries (Pandas, NumPy, etc.)
• OR equivalent experience.
• 2+ years of experience in data analysis or data engineering, including work with large-scale datasets that are unstructured or semi-structured.
• Proficiency in statistics and exploratory data analysis methods.
• Familiarity with data processing frameworks such as Spark, Ray, or Apache Beam.
• Ability to communicate technical findings clearly to research and product teams.