SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Replit is seeking an AI Agent Security Architect to serve as the primary technical authority for securing autonomous AI agents on its platform. This is a high-impact individual contributor role focused on designing and implementing runtime defense systems, guardrail frameworks, and sandboxing architectures that govern AI agents executing code and reasoning across the platform.
Key responsibilities include:
**AI Agent Security Strategy & Technical Execution**
Define long-term architectural patterns for securing autonomous agent workflows, Model Context Protocol (MCP) integrations, multi-turn reasoning loops, and multi-agent coordination. Architect and deploy dynamic input/output guardrail systems, semantic firewalls, and real-time intent verification filters to prevent goal hijacking, system prompt leaks, and indirect prompt injections.
**Agent Execution & Tool Sandboxing**
Partner with Infrastructure and AppSec teams to design secure, short-lived, micro-isolated environments (microVMs, WebAssembly, container sandboxes) where agents can dynamically execute code and interact with host operating systems safely.
**Agentic Threat Modeling & Red Teaming**
Conduct specialized threat modeling against non-deterministic systems. Lead automated and manual AI red-teaming initiatives to uncover vulnerabilities in RAG context pipelines, vector stores, and tool-calling interfaces.
**Identity, Delegation & Least Privilege**
Architect fine-grained permission models and Human-in-the-Loop (HITL) authorization patterns for agents acting on behalf of users, ensuring agents never exceed delegated privileges or misuse API keys.
**Context & Memory Boundary Security**
Design strict data isolation and poisoning prevention controls for vector databases, RAG pipelines, and long-term conversation memories across multi-tenant workloads.
**Risk Management & Cross-Functional Enablement**
Maintain authoritative secure design patterns and SDKs for engineering teams. Identify and document non-deterministic and agentic security risks in the Cybersecurity Risk Register. Design tamper-evident logging and telemetry standards for agent reasoning chains and tool execution history. Partner with GRC to translate AI security controls into enterprise-ready audit documentation (ISO 42001, NIST AI RMF, EU AI Act readiness) and support Sales on complex enterprise security inquiries.
**Required Experience:**
6+ years in security engineering or architecture, with 2+ years dedicated to AI/ML security, LLM application security, or securing agentic frameworks. Deep understanding of agentic architectures (MCP, LangChain, AutoGen, CrewAI, ReAct frameworks, vector databases). Hands-on expertise with agentic threat vectors: indirect prompt injection, goal hijacking, tool parameter tampering, data poisoning, context contamination. Strong background in OWASP Top 10 for LLMs & AI Agents, NIST AI RMF, and MITRE ATLAS. Proficiency in Python, Go, or TypeScript/JavaScript with proven track record of reviewing code and building security tooling. Experience authoring and evangelizing technical security architecture documentation. Exceptional ability to communicate complex AI risks to both engineers and executives. Proven contribution to enterprise Cybersecurity Risk Registers.
**Bonus:** Experience designing runtime sandboxing and isolation technologies (Firecracker, gVisor, Docker, WebAssembly) for dynamic code execution environments.
