Multimodal Safety
Core
Develop model-based classifiers, runtime guardrails, evaluations, and data to ensure generative-media models (image generation/editing) produce safe, creative, and useful content while respecting safety policies.
Role type
Member of Technical Staff (ML Engineer)
Builds
Runtime safeguards, policy models, evaluation datasets, and monitoring infrastructure for multimodal AI systems.
Domain
Generative AI, Multimodal Systems, Content Safety
Deliverable
production ML models
Required skills
Machine learning system development, Python programming, experimental design and metrics, data pipeline construction, reproducible ML workflows, cross-functional collaboration
Preferred skills
Diffusion and autoregressive media models, image/audio safety classifiers, adversarial evaluation and red teaming, Azure cloud platform experience, responsible AI policy development
Technologies
Python, Azure, Azure AI services, Azure Machine Learning, GPU capacity planning
Responsibilities
Develop learned classifiers and prompt-based safeguards; Build training and evaluation datasets from synthetic generation and real-world traffic; Design evaluations for harmful or deceptive generated media; Diagnose failures and translate findings into model or policy improvements; Monitor deployed safeguards across traffic sources and product configurations; Partner with production teams on Azure-based deployments and guardrails integration.
Seniority
Mid-level IC (4+ years experience)
