Multimodal Data Engineer

GaussMathematics is building a world data infrastructure for models that understand the world.

We work with virtual worlds, simulation systems, and physical environments to produce structured multimodal data engines for world models, robotics, and physical AI. Our work includes world-state snapshots, synchronized multi-view video, action-state trajectories, replayable traces, benchmark tasks, evaluation records, and customer-specific data pipelines.

We work with frontier teams building the next generation of spatial intelligence, embodied AI, robotics systems, generative video, simulation, and physical AI.

We are looking for a Multimodal Data Engineer to build synchronized data systems across video, state, actions, metadata, simulation records, and evaluation signals.

This is an entrepreneurial, high-ownership engineering role at the intersection of data engineering, computer vision, ML data workflows, and multimodal AI. You will work on systems that align, validate, transform, and package heterogeneous signals into coherent training and evaluation records.

This is not a conventional data engineering role focused only on tables or batch pipelines. The work is about building structured multimodal records that preserve time, state, interaction, and context.

You will help define how complex multimodal environments become model-ready data.

• Build pipelines for synchronizing and processing video, images, metadata, state records, actions, sensor streams, and evaluation outputs.
• Design data schemas and transformation workflows for multimodal training and evaluation records.
• Build tools for alignment, validation, inspection, packaging, and versioning.
• Work closely with infrastructure, simulation, robotics, ML, and customer delivery teams.
• Help turn raw multimodal signals into coherent datasets and customer-specific data engines.
• Improve quality, traceability, reproducibility, and observability across data workflows.
• Debug complex data issues across formats, timestamps, missing fields, inconsistent metadata, and customer-specific environments.
• Build systems that can support frontier model training, post-training, evaluation, and review loops.

• Experience with data engineering, ML data pipelines, computer vision data, multimodal datasets, or data infrastructure.
• Strong ability to work with messy, heterogeneous, high-volume data.
• Familiarity with video, images, metadata, annotations, state records, logs, sensors, or evaluation outputs.
• Comfort designing schemas, validation logic, transformation pipelines, and inspection tools.
• Strong debugging skills and attention to detail.
• Ability to work in ambiguous, customer-specific technical environments.
• A builder mindset: you are excited to create reliable systems from complex raw signals.

• Experience with Python, SQL, TypeScript, Spark, Ray, workflow engines, data warehouses, object storage, or similar tools.
• Familiarity with computer vision datasets, video pipelines, annotation systems, ML training data, robotics data, or simulation records.
• Experience building dataset tooling, QA systems, data validation frameworks, internal platforms, or ML infrastructure.
• Experience with time synchronization, multimodal alignment, metadata management, or evaluation pipelines.
• Experience in an early-stage company, AI lab, data platform team, or 0–1 technical environment.

Success in this role means building the systems that turn heterogeneous multimodal signals into coherent, reliable, model-ready data.

You will help GaussMathematics create the data foundation for models that need to understand environments, actions, and outcomes.

GaussMathematics is working on a foundational problem: how to turn complex environments into structured data systems for models that need to understand, predict, and act in the world.

This is an opportunity to work at the center of multimodal AI, data infrastructure, and world model development.

Send a note to:

walkingtogether@gaussmathematics.ai

Please include your background, relevant experience, examples of multimodal, ML data, data infrastructure, or technical systems you have built, and anything else that helps us understand what makes you different.