Skip to main content
TALENT247 // SPECIALIST AI EXECUTIVE SEARCH & TECHNICAL RECRUITMENT
APP · Applied AI

Applied AI Security & Guardrails Architect

New York, NY · Full-time · Hybrid

AI Specialism

Applied AI

Workplace

Hybrid

Location

New York, NY

Compensation

$255k – $345k

annual package

Role Dossier & Specification

REF: APPLIED-AI-SECURIT

THE CLIENT & MANDATE

Our client is a leading safety-focused frontier AI company headquartered in San Francisco with a rapidly expanding enterprise and security engineering presence in New York City. As global banks, sovereign institutions, healthcare networks, and software platforms connect autonomous coding and knowledge agents directly to internal production infrastructure via Model Context Protocol (MCP) and computer-use runtimes, defense-in-depth agent security has become paramount.

They have partnered with Talent247 to appoint an Applied AI Security & Guardrails Architect in New York, NY. Sitting across security engineering, applied AI, and strategic enterprise architecture, the successful candidate will design zero-trust agent sandboxes, indirect-prompt-injection defenses, and cryptographic capability-control frameworks protecting the world's most security-sensitive AI deployments.

KEY RESPONSIBILITIES

  • Architect defense-in-depth runtime security patterns for tool-using and computer-use agents—spanning capability-scoped MCP gateways, microVM/network egress sandboxing, and deterministic data-exfiltration controls.
  • Design layered ML + deterministic guardrail pipelines mitigating indirect prompt injection across untrusted web pages, emails, code repositories, and third-party SaaS records.
  • Partner directly with CISO and Principal Security Architects at flagship financial services, life sciences, and technology customers to threat-model multi-agent production rollouts.
  • Build automated continuous red-teaming harnesses that simulate confused-deputy attacks, credential harvesting, MCP tool poisoning, and cross-tenant memory leakage.
  • Co-design enterprise governance primitives with core platform teams—including cryptographic audit trails, zero-data-retention enclaves, customer-managed KMS keys, and policy-as-code tool approvals.
  • Publish authoritative reference architectures and open security standards for safe enterprise agent deployment.

TECHNICAL ENVIRONMENT & OFFERING

Stack: Python, Rust, TypeScript, Go, Model Context Protocol (MCP), Firecracker/gVisor/seccomp-bpf, Envoy/mTLS, OAuth 2.1 / OIDC, cloud KMS/confidential computing, frontier LLM classifier guardrails. Workplace: Hybrid in New York, NY. Compensation: $255,000 – $345,000 base salary plus high-value frontier lab equity and comprehensive US benefits.

Core Requirements

ESSENTIAL QUALIFICATIONS

  • 7+ years of security architecture, application/cloud security engineering, or secure distributed systems development, including 2+ years focused on LLM/agent security.
  • Deep technical understanding of AI-native attack surfaces: indirect prompt injection, tool/MCP schema poisoning, sandbox escape, data exfiltration side-channels, and jailbreak chaining.
  • Strong systems and coding fluency (Python, Go, Rust, or TypeScript) capable of building working reference sandboxes, policy proxies, and adversarial exploit harnesses.
  • Mastery of enterprise cloud security controls (AWS/GCP/Azure IAM, VPC isolation, OAuth2/OIDC token scoping, zero-trust networking, confidential computing).
  • Clear, pragmatic communication style trusted equally by internal alignment researchers and Fortune 100 CISOs.

PREFERRED / BONUS EXPERIENCE

  • Prior offensive security research, bug-bounty discoveries, or published research on securing tool-using AI agents.
  • Experience securing developer IDE/CLI coding agents or enterprise SaaS connector platforms at scale.
  • Contributions to OWASP GenAI, MCP security specifications, or open-source sandboxing projects.

Position Telemetry

Location New York, NY (Hybrid)
Department Applied AI
Employment type Full-time
Workplace type Hybrid
Salary range
$255k – $345k annual package