AI Agent Security Architect

Location: Hybrid in Foster City, California, United States.

Compensation: $230–$350, plus commission offers.

Company

Replit is an agentic software creation platform that enables users to build applications with natural language.

What you will do

• Define the security blueprint and architectural patterns for autonomous agent workflows, MCP integrations, multi-turn reasoning, and multi-agent coordination.
• Architect runtime guardrails, semantic firewalls, intent verification filters, and policy enforcement systems to prevent goal hijacking, prompt leaks, and indirect prompt injection.
• Design secure short-lived, micro-isolated execution environments for agents running code, shell commands, and tool interactions.
• Lead agentic threat modeling and AI red-teaming across RAG pipelines, vector stores, context assembly, and tool-calling interfaces.
• Build identity, delegation, least-privilege, and human-in-the-loop authorization controls, along with data isolation and poisoning prevention for multi-tenant workloads.
• Maintain secure design patterns and SDKs, contribute to the cybersecurity risk register, define tamper-evident observability standards, and support compliance documentation and enterprise security inquiries.

Requirements

• 6+ years of experience in security engineering or security architecture, including at least 2+ years dedicated to AI/ML security, LLM application security, or agentic framework security.
• Deep understanding of agentic architectures and protocols, including MCP, LangChain, AutoGen, CrewAI, ReAct frameworks, and vector database technologies.
• Hands-on expertise with indirect prompt injection, goal hijacking, tool parameter tampering, data poisoning, and context contamination.
• Strong background in OWASP Top 10 for LLMs and AI Agents, NIST AI RMF, and MITRE ATLAS.
• Proficiency in Python, Go, or TypeScript/JavaScript, with experience reviewing code and building functional security tooling or guardrails.
• Experience authoring, maintaining, and evangelizing technical security architecture documentation.
• Ability to communicate complex non-deterministic AI risks to technical engineers and executive stakeholders.
• Proven experience contributing to an enterprise Cybersecurity Risk Register.

Nice to have

• Experience designing runtime sandboxing and isolation technologies such as Firecracker, gVisor, Docker, or WebAssembly for dynamic code execution environments.

Culture & Benefits

• Autonomous work environment with a hybrid office setup.
• Competitive salary and equity.
• 401(k) program with a 4% match for US employees.
• Health, dental, vision, life, disability, parental, medical, and caregiver benefits.
• Flexible time off, holidays, monthly wellness stipend, commuter benefits, office setup reimbursement, office amenities, and quarterly team gatherings.