Safety Evaluation Engineer: Turn AI Risk into Metrics

Meta; Menlo Park (CA); On-site; USD 219,000 - 301,000; Full time

Benefits offered by this job

Bonus potential

Equity

Benefits

Job summary

Meta is seeking a Research Engineer to join the Safety Evaluation team within Meta Superintelligence Labs. The role focuses on designing evaluations that measure safety across text, image, voice, video, and agentic systems, and building the infrastructure to run them at scale inside training and production environments.

You will set measurement standards, curate safety datasets, and drive the direction of safety metrics.

Qualifications

• Bachelor's degree in CS/Engineering or related field
• 3+ years of ML/AI research or research-engineering experience in ML/AI including hands-on work with LLMs, multimodal models, or NLP
• Experience designing and validating evaluations or benchmarks for ML systems
• Experience building production-grade or research infrastructure that must be reliable at scale
• Programming in Python and hands-on experience with PyTorch
• Ability to communicate complex technical results to non-specialist stakeholders and decision-makers

Responsibilities

• Set the technical strategy for safety evaluation across model families and modalities
• Design and validate novel evaluations for safety-critical behaviors
• Build and harden the distributed evaluation platform for large-scale training runs
• Own the measurement quality bar and determine when an eval is trustworthy to gate a launch
• Create, curate, and analyze safety datasets including multilingual and long-tail cases
• Translate evolving global safety policy into concrete measurement criteria
• Mentor engineers and researchers and raise the evaluation bar across the org
• Represent safety evaluation methodology to leadership and external audiences

Skills

Python

PyTorch

ML/AI research

Multimodal models

NLP

Communication

Stakeholder management

Adversarial evaluation

Education

Bachelor's degree in CS/Engineering

Job description

Meta is seeking a Research Engineer to join the Safety Evaluation team within Meta Superintelligence Labs. The role focuses on designing evaluations that measure safety across text, image, voice, video, and agentic systems, and building the infrastructure to run them at scale inside training and production environments.

You will set measurement standards, curate safety datasets, and drive the direction of safety metrics.