AI Agent Security Architect
Design and lead security architecture for Replit’s autonomous AI agents, including runtime guardrails, sandboxing, threat modeling, least-privilege delegation, and observability. Requires 6+ years in security engineering or architecture and specialized experience securing AI/ML or agentic systems.
About the job
Responsibilities
- Define Replit’s long-term security architecture for autonomous agent workflows, Model Context Protocol (MCP) integrations, multi-turn reasoning loops, and multi-agent coordination.
- Architect runtime guardrails, semantic firewalls, real-time intent verification, and policy enforcement 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 tools.
- Conduct agentic threat modeling and lead automated and manual AI red-teaming across RAG pipelines, vector stores, and tool-calling interfaces.
- Build fine-grained permission, delegation, and Human-in-the-Loop authorization patterns based on least privilege.
- Design data isolation and poisoning-prevention controls for vector databases, RAG pipelines, and long-term memories in multi-tenant workloads.
- Maintain authoritative AI security design patterns and SDKs for engineering teams.
- Identify and document agentic security risks in the cybersecurity risk register.
- Establish tamper-evident logging and telemetry for agent reasoning chains, tool execution, and context assembly.
- Support GRC and Sales with AI security controls, audit documentation, and enterprise security inquiries.
Requirements
- 6+ years of security engineering or security architecture experience, including at least 2 years focused on AI/ML security, LLM application security, or agentic frameworks.
- Deep understanding of agentic architectures and protocols, including MCP, LangChain, AutoGen, CrewAI, ReAct, and vector databases.
- Hands-on experience with indirect prompt injection, goal hijacking, tool parameter tampering, data poisoning, and context contamination.
- Strong knowledge of 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 security tooling or guardrails.
- Experience authoring and maintaining technical security architecture documentation.
- Ability to communicate complex AI risks to technical and executive audiences.
- Experience contributing to an enterprise cybersecurity risk register.
Nice-to-haves
- Experience designing runtime sandboxing and isolation technologies such as Firecracker, gVisor, Docker, or WebAssembly.
- Familiarity with ISO 42001 and EU AI Act readiness.
Compensation and Benefits
- Salary and equity.
- 401(k) program with a 4% match for US employees.
- Health, dental, vision, and life insurance.
- Short- and long-term disability coverage.
- Paid parental, medical, and caregiver leave.
- Flexible time off and holidays.
- Commuter benefits for in-office US employees.
- Monthly wellness stipend.
- Autonomous work environment.
- In-office setup reimbursement.
- Quarterly team gatherings and in-office amenities.
Skills
Ai Security, Security Architecture, Model Context Protocol, LangChain, Autogen, Crewai, React, Vector Databases, Python, Go, TypeScript, Webassembly, Docker, Firecracker, Gvisor
Similar jobs
Security Engineering jobs
Coalition Security
United StatesInvestigates cyber incidents and insurance claims, assesses security controls, and advises customers and security leaders on prioritized risk improvements. The role requires 2–4 years of security experience, strong network threat knowledge, and familiarity with major security and compliance frameworks.
Idme
McLean, VA
Leads cloud-native security operations, incident response, threat hunting, and forensic investigations while mentoring SOC analysts and improving detection processes. Requires 8+ years in information security, including hands-on cloud incident response and experience with Kubernetes, CI/CD, and advanced security tools.
ConductorOne
San Francisco, CASenior Security Engineer responsible for application, cloud, and platform security, with a focus on automating security workflows, threat modeling, secure development, and remediation. Requires hands-on SaaS security, cloud infrastructure, code review, and agent or automation experience.
BlackCloak
United States
Leads end-to-end response to sophisticated compromises, account takeovers, device threats, and related financial fraud for high-profile clients. Requires 5–8+ years of incident response or DFIR experience, broad device forensics expertise, strong client communication, and fraud-remediation experience.
Supernova Technology
Chicago, IL
The Senior Information Security Engineer will lead threat hunting, detection engineering, incident response, vulnerability management, and security-platform ownership across cloud and enterprise environments. The role requires 5+ years of security experience, strong attacker-TTP knowledge, and hands-on expertise with enterprise security technologies and automation.