Member of Technical Staff - Imagine Safety

Written by the T&S Network from SpaceXAI's posting.

Product safety engineering for Grok's multimodal image, video and audio generation on the Imagine team. The work is building the safeguards, detection systems and evaluation frameworks that prevent harm in generated media while keeping creative use open. The role sits within the Imagine team, which is redefining AI-driven media experiences for Grok users.

• AI misuse and jailbreaks

What you'd do

• Design and implement scalable real-time moderation, risk detection and safeguard enforcement for images, video and audio.
• Build infrastructure to measure and mitigate harmful content, bias, deepfakes, IP issues and misuse at global scale.
• Develop tools, pipelines and evaluation frameworks that let safety policies iterate quickly with research, product and policy.
• Use feedback from user interactions, model outputs and training data to keep tightening safety without adding latency.
• Own safety features end to end, from problem definition and prototyping through deployment, monitoring, incident response and refinement.
• Embed safety into core Imagine features with the engineering, research and product teams.

What it asks for

• Proficiency in Rust with a track record of clean, scalable, maintainable code.
• Years building safety, trust and safety, or moderation systems for consumer products at large scale.
• Direct work on live detection, data pipelines or evaluation tooling for AI systems handling heavy traffic.
• Proof of shipping dependable systems used by millions with strong uptime.
• Experience with multimodal content safety for images, video or audio, or generative AI safety in production (preferred).
• Familiarity with ML classifiers, safety evaluation, red-teaming or adversarial testing for media generation models (preferred).
• Experience with distributed systems, live model inference, Kubernetes or observability stacks (preferred).

Terms in this role

Red teaming.

Testing a system, often an AI model, by deliberately trying to make it fail or misbehave, so problems are found and fixed first.

Our best summary of the posting, and it can be wrong. The employer's posting is what counts: read it in full on SpaceXAInorth_east before you apply. If we got something wrong, report this listing.