Multimodal Data Engineer - Video & Audio Pipelines

Doist; Palo Alto (CA); On-site; USD 120,000 - 180,000; Full time

Job summary

Odyssey is an AI lab building general world models. We seek a data engineer to build and curate large-scale multimodal datasets across video, robotics, and audio, working closely with researchers and infrastructure teams.

You’ll balance data quality, extract signals, and tune pipelines for training at scale. The role blends research focus with practical platform work, requiring hands-on data experience with video and audio data.

Qualifications

• Experience building data pipelines at scale across multimodal datasets.
• Hands-on with video and audio data processing for ML workloads.
• Ability to switch between data engineering and research-facing work.

Responsibilities

• Build and run data pipelines for large-scale multimodal datasets (video, robotics, audio).
• Extract useful signals from raw data (speakers, background audio, features in video).
• Balance and deduplicate data, optimize training datasets and signals.
• Collaborate with researchers to translate model requirements into data strategies.
• Improve storage, databases, throughput, and deployment for reliability.

Skills

Data engineering

Data pipelines

Multimodal data

Video data

Audio data

Research collaboration

Platform & infrastructure

Data analysis

Tools

SQL

Python

Spark

GCS/S3

Job description

Odyssey is an AI lab building general world models. We seek a data engineer to build and curate large-scale multimodal datasets across video, robotics, and audio, working closely with researchers and infrastructure teams.

You’ll balance data quality, extract signals, and tune pipelines for training at scale. The role blends research focus with practical platform work, requiring hands-on data experience with video and audio data.