Generative Media Quality Engineer

Meta Careers; San Francisco (CA); On-site; USD 131,000 - 190,000; Full time

Job summary

Meta seeks a Product Content Engineer to own evaluation rubrics and human assessments for image and video generation. You will design rubrics, run evaluation rounds, and produce writeups that translate subjective quality into measurable signals guiding model behavior and launches.

You will curate test prompts and golden sets, calibrate with raters and vendors, and help bridge human judgment with automated metrics in a fast-paced environment.

Qualifications

• Hands-on video production experience and craft vocabulary to describe quality failures precisely.
• Experience creating or contributing to evaluation rubrics or quality guidelines for video or image content.
• Experience producing or evaluating short-form video and emerging video formats.
• Ability to document judgment logic into reusable scalable frameworks.

Responsibilities

• Build and maintain evaluation rubrics for image and video generation quality, including definitions and guidelines.
• Execute human evaluation rounds on model outputs and produce actionable writeups.
• Construct and curate prompt sets and golden sets to stress-test model capabilities.
• Calibrate with raters and vendor teams to ensure reliable assessments.
• Support translation of human judgment into automated measurement and validate automated scores against human ground truth.

Skills

Video production

Editorial standards

Cross-functional collaboration

Quantitative insights

Independent work

Education

Bachelor's degree in a related field

Tools

Annotation platforms

Eval tooling

Job description

Meta seeks a Product Content Engineer to own evaluation rubrics and human assessments for image and video generation. You will design rubrics, run evaluation rounds, and produce writeups that translate subjective quality into measurable signals guiding model behavior and launches.

You will curate test prompts and golden sets, calibrate with raters and vendors, and help bridge human judgment with automated metrics in a fast-paced environment.