Senior Staff AI Security Lead

IonQ

10d•$180,000 - $225,000•Remote

About The Position

We are hiring a senior technical leader to build the AI foundation our security organization runs on — and to make sure it actually gets used. This is a dual-mandate role. You will stand up the enterprise AI infrastructure that lets security teams safely build with large language models and agents, and you will lead the adoption effort that turns that infrastructure into measurable gains across detection and response, threat intelligence, vulnerability management, GRC, and security engineering. This is a hands-on senior individual contributor role. You will write code, design systems, and set technical direction. You will not manage a team, but you will lead through influence: partnering with engineering leaders, coaching analysts and engineers, and building the coalitions that make org-wide change stick. Expect roughly an even split between platform building and enablement work. You will report to the security leadership team and work closely with Platform Engineering, IT, Legal, Privacy, and the broader AI/ML organization.

Requirements

8+ years in security engineering, platform engineering, or a closely related technical field, including significant hands-on software development.
Demonstrated experience building and operating production systems with large language models — not prototypes or demos. You have shipped something real, dealt with its failure modes, and iterated on it.
Strong software engineering fundamentals and fluency in Python (or an equivalent language), including API design, distributed systems, and cloud infrastructure (AWS, GCP, or Azure).
Deep understanding of security engineering: identity and access management, secrets management, network boundaries, logging and detection, and secure software development practices.
Working knowledge of AI-specific risk — prompt injection, jailbreaks, data leakage through model outputs, insecure agent tool use, model and dependency supply chain — and practical mitigations for each.
A track record of driving technical change across an organization through influence rather than authority: building consensus, teaching, and shipping things people voluntarily adopt.
Clear written and verbal communication. You can explain a system design to an engineer and the business case for it to an executive, in the same week.
Sound judgment about where AI is genuinely useful and where it is not. We want an advocate who is also a skeptic.

Nice To Haves

Experience with agent frameworks, orchestration, and the Model Context Protocol (MCP) or comparable tool-use standards.
Experience building evaluation frameworks or LLM observability tooling.
Background in security operations, detection engineering, or incident response — you have felt the toil firsthand.
Experience with retrieval systems, embeddings, vector databases, and knowledge pipeline design.
Familiarity with AI governance and assurance frameworks such as the NIST AI Risk Management Framework, ISO/IEC 42001, or the OWASP Top 10 for LLM Applications.
Experience fine-tuning or evaluating open-weight models for domain-specific tasks.
Prior experience as a first or founding hire on a platform or capability that later scaled organization-wide.

Responsibilities

Stand Up the Enterprise AI Infrastructure: Build the platform. Design and build the shared platform security teams use to develop AI applications — model gateway and routing, authentication and authorization, secrets handling, rate limiting, cost attribution, and audit logging.
Enable agentic systems. Establish patterns and reusable components for agentic workflows: tool and MCP server integration, sandboxed execution, human-in-the-loop approval gates, and least-privilege scoping for agent credentials.
Wire in the data. Connect security's knowledge — runbooks, detection logic, past incidents, asset inventory, policy documents — to AI systems through well-governed retrieval pipelines with correct access controls and data classification enforcement.
Make it measurable. Build the evaluation harness, regression suites, and observability that tell us whether an AI system is working: output quality, latency, cost, hallucination rate, and drift over time. Nothing ships to production without a baseline.
Secure the stack itself. Implement guardrails against prompt injection, data exfiltration through model outputs, insecure tool use, and supply-chain risk in models and AI dependencies. Red-team the platform and the applications built on it.
Set the standards. Own the reference architecture, golden paths, and internal documentation so that a security engineer can go from idea to reviewed prototype in days rather than months.
Drive AI Adoption Across the Security Organization: Find the real use cases. Work directly with detection engineering, IR, threat intel, vulnerability management, GRC, and security operations to identify where AI meaningfully reduces toil or improves quality — and where it does not. Kill weak ideas early.
Build the flagship applications. Deliver a small number of high-visibility wins yourself. Nothing drives adoption like a working tool that saves an analyst two hours a day.
Teach the organization. Run enablement programs — office hours, workshops, internal documentation, prompt and agent design patterns, brown-bags — that raise the AI fluency of the entire security organization, not just the engineers.
Measure adoption honestly. Define and report the metrics that show whether adoption is real: active users, workflows automated, analyst hours returned, time-to-detect and time-to-respond improvements, quality deltas against human baselines.
Establish governance that enables. Partner with Legal, Privacy, and Compliance to establish the acceptable-use policy, review process, and risk framework for AI in security workflows — designed to unblock teams rather than stall them.
Stay ahead of the field. Maintain the organization's point of view on where AI capability is heading and what it means for our security roadmap, staffing, and threat model. Evaluate vendors and open models on the merits.