Replit is the agentic software creation platform that enables anyone to build applications using natural language. With millions of users worldwide, Replit is democratizing software development by removing traditional barriers to application creation.
The Trust & Safety team is the front line defending Replit's platform from exploitation. We detect and shut down phishing deployments, prevent cryptomining on free-tier infrastructure, stop LLM token farming, and keep bad actors from weaponizing the platform against our users. This is adversarial work: attackers adapt constantly, and we build the detection systems, heuristics, and automated responses that stay ahead of them.
What makes this role unique is the AI-native nature of Replit's platform. You'll work on problems that barely exist elsewhere: building guardrails for AI-generated code, detecting prompt injection attacks at scale, and using LLMs as a defensive tool against abuse. If you want hands-on experience applying AI to security problems, this is one of the few places you can do it in production with real attackers. You'll own problems end-to-end, from identifying emerging abuse patterns to shipping the systems that stop them at scale.
עיקרי התפקיד
Design and implement LLM guardrails that detect abuse scenarios in AI-generated code and agent interactions
Build AI-powered detection systems that use LLMs to identify malicious patterns, classify threats, and automate response decisions
Build and operate abuse detection systems that identify phishing, cryptomining, account takeover, and financial fraud across millions of daily user actions
Reasons not to work at Replit https://blog.replit.com/reasons-not-to-join-replit
To achieve our mission of making programming more accessible around the world, we need our team to be representative of the world. We welcome your unique perspective and experiences in shaping this product. We encourage people from all kinds of backgrounds to apply, including and especially candidates from underrepresented and non-traditional backgrounds.