Multimodal AI Engineer - Audio, Video & Data Pipelines

Meta; Lincoln (NE); On-site; USD 154,000 - 217,000; Full time

Job summary

Meta’s Applied AI (AAI) organization leads multimodal research and production systems. The Multimedia & Modality team covers image, video, audio, speech and music, collaborating with research, model-training and engineering groups.

As a Software Engineer in Multimedia & Modality AI, you will design data pipelines, evaluation methods, and agentic workflows, shaping how we measure and improve models across all media modalities.

Qualifications

• 3+ years building ML systems in production or research settings.
• Strong Python and PyTorch experience.
• Experience with speech, audio, or music ML is desirable.
• Experience with large-scale data pipelines and distributed training.
• Track record of translating research ideas into working systems.

Responsibilities

• Design and implement agentic workflows and pipelines, including human-in-the-loop designs.
• Design and own data pipelines at scale: ingestion, filtering, pseudo-labeling and captioning with attribute classifiers.
• Build evaluation infrastructure: objective metrics, listening tests, and correlation analyses.
• Improve training efficiency and reliability: distributed training, GPU utilization, tokenizer retraining, experiment management.
• Reproduce and extend state-of-the-art research by implementing methods from papers into codebase and running ablations.
• Mentor engineers, contribute to hiring and onboarding, and raise the bar on evaluation and quality practices.
• Build and train generative and representation models for speech, sound, and music.

Skills

Python

PyTorch

ML systems

Distributed training

Data pipelines

Job description

Meta’s Applied AI (AAI) organization leads multimodal research and production systems. The Multimedia & Modality team covers image, video, audio, speech and music, collaborating with research, model-training and engineering groups.

As a Software Engineer in Multimedia & Modality AI, you will design data pipelines, evaluation methods, and agentic workflows, shaping how we measure and improve models across all media modalities.