Member of Technical Staff, Trust & Safety Engineer

What you’ll do

• Serve as a core safety partner embedded across product and research teams, providing Trust & Safety engineering support for all launches - from early design review through post-launch monitoring
• Build and maintain safety infrastructure that ensures Runway's models have a positive impact on the world as they reach millions of users
• Design, execute, and continuously improve red teaming systems to proactively surface model safety issues, such as harmful content generation, filter bypasses, and policy-violating outputs, before they reach production
• Translate ambiguous, fast-evolving T&S challenges into concrete technical solutions, often without an existing playbook; comfort with navigating uncertainty and proposing novel approaches is essential
• Improve the performance, reliability, and code quality of our codebases

What you’ll need

• 3+ years of experience in software engineering, with production experience in Python or TypeScript
• A proactive, ownership-oriented mindset - you'll own systems end-to-end (infrastructure, monitoring, logging, metrics, incident response) and be responsible for identifying risks, not just responding to tickets
• Comfort working across the stack - from writing model evaluation scripts to building internal tooling to debugging infrastructure issues in AWS or GCP
• Experience with data pipelines or analytics infrastructure (we use ClickHouse heavily for abuse detection and policy enforcement signals)
• Ability to translate ambiguous, fast-evolving policy requirements into concrete technical solutions - you'll often be defining the approach, not following an existing playbook
• Strong written communication skills; you'll work closely with legal, policy, and product teams to document safety decisions and explain technical tradeoffs
• Humility and open mindedness; at Runway we love to learn from one another

Technical Stack

• API for content moderation and policy enforcement written in TypeScript, running in ECS containers on AWS Fargate
• AWS-native components: S3, CloudFront, Lambda, Kinesis, SQS
• Inference backend written in Python (PyTorch), deployed across multiple clusters
• Leverage LLMs (Anthropic Claude, Google Gemini) as part of safety and moderation systems
• Prometheus and Grafana for monitoring
• Terraform for infrastructure management