SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Salary: USD 190,000 - 300,000 / annual
Thinking Machines Lab is seeking an Operations Analyst focused on Safety to help build trustworthy AI products. The role combines hands-on content moderation with strategic policy development and tooling automation.
Day-to-day responsibilities include reviewing flagged content, safety escalations, and abuse signals in a standing queue (not rotational). You'll triage cases, apply policy judgment, and take enforcement actions including content flags, account reviews, and bans/recovery decisions. This casework directly informs your strategic work.
Beyond the queue, you'll design and refine safety policies across the product stack in collaboration with engineering, legal, safety research, and security teams. You'll build and maintain tooling and automation—triage agents, ban/recovery workflows, abuse detection frameworks—that make moderation faster and more consistent. You'll partner with product teams to embed safety into the user experience through model refusals, content flagging, account review features, and safety protections informed by real queue patterns.
You'll also improve observability and detection for safety-relevant events, surfacing abuse patterns and malicious behavior in production so cases surface faster with better signal.
Minimum qualifications: 2+ years in trust & safety, content moderation, or fraud/abuse operations with direct recurring queue responsibility. Proven experience owning policy definition, operationalization, and enforcement end-to-end. Direct case experience with cybersecurity abuse, CBRN-relevant risk, youth safety, or prompt injection in production. Working familiarity with model safety and abuse risk categories (jailbreaks, prompt injection, scaled abuse). Practical experience using AI tools (Claude, Codex, or similar) to build or accelerate operational workflows.
Preferred: Safety and integrity operations on AI-powered products or LLM APIs; track record of turning case patterns into reusable tooling while owning the underlying queue; experience training or setting quality bars for other moderators.
You'll thrive if you have thoughtful opinions about safe, trustworthy user experiences grounded in real cases; can translate safety constraints into product trade-offs; and bias toward speed and learning measured by queue improvements.
