Multimodal Data Engineer — Video & Audio Pipelines

This is Growth; San Francisco (CA); On-site; USD 120,000 - 210,000; Full time

Job summary

This is Growth seeks a data engineer specializing in multimodal ML data. You will build and curate large training corpora across video, robotics and audio, ensuring alignment with model objectives and quality standards.

You’ll partner with researchers to tune caption quality, manage dataset versioning, and scale pipelines that process petabytes of data cost-effectively. The role blends hands-on data engineering with research collaboration in a fast-moving AI lab.

Qualifications

• Built training corpora for visual or multimodal ML at scale and wrote code to judge quality.
• Hands-on with video and audio — codecs, decoding, frame-level work, keeping modalities in sync.
• Proficient Python, Kubernetes in production, Ray or Flyte, petabyte-scale object storage.
• Able to justify data pruning decisions.

Responsibilities

• Curate large-scale multimodal datasets across video, robotics and audio.
• Define selection logic: filtering, deduplication, quality scoring, captioning and shot detection.
• Platforms processing millions of hours of content cost-effectively.
• Collaborate with researchers on caption quality and data mix.

Skills

Multimodal ML

Python

Kubernetes

Ray

Flyte

Data lineage

Job description

This is Growth seeks a data engineer specializing in multimodal ML data. You will build and curate large training corpora across video, robotics and audio, ensuring alignment with model objectives and quality standards.

You’ll partner with researchers to tune caption quality, manage dataset versioning, and scale pipelines that process petabytes of data cost-effectively. The role blends hands-on data engineering with research collaboration in a fast-moving AI lab.