Agentic AI Security Engineer

Howard Hughes Medical Institute; Ashburn (VA); On-site; USD 149,000 - 264,000; Full time

Job summary

Howard Hughes Medical Institute is seeking a security engineer to design and operate the safetystack for agentic AI at Janelia, ensuring safe execution of agent code on OpenShift and HPC platforms.

You will implement sandboxing, guardrails, and full observability with auditable sessions, while contributing to foundational architecture decisions and open-sourcing efforts.

Qualifications

• Hands-on experience operating LLM agents and dealing with their failures or deep sandbox/container-security expertise.
• Linux and Kubernetes/OpenShift isolation depth with precise understanding of containerization vs sandboxing.
• Knowledge of agent architectures: tool-calling protocols, approval gates, judge models, guardrail frameworks.
• Open-source contributions, publications, or production systems you can discuss in depth.
• Minimum 5 years in security, systems/SRE, or ML systems engineering.

Responsibilities

• Research, develop, deploy, operate, and analyze the safetystack for agentic AI at Janelia.
• Design sandboxed, reproducible execution environments with kernel- and VM-level isolation.
• Implement guardrails and observability, with audit logging and post-incident analysis.

Skills

Sandboxing

Security engineering

LLM agent experience

Kubernetes/OpenShift

Adversarial testing

Observability

Education

Bachelor's degree in a relevant field

Tools

OpenShift

Kubernetes

Linux

HPC cluster

Job description

Primary Work Address: 19700 Helix Drive, Ashburn, VA, 20147

AI@HHMI at Janelia integrates AI agents across the researchprocess. These tools write and execute code, optimize experimentaland model parameters, and drive instruments.

Letting AI agents act on shared research infrastructure requiresserious safety engineering. The relevant failure mode is usuallynot a malicious attacker but a capable, misdirected tool runningwith legitimate user permissions at machine speed: hallucinatedcommands, instructions hijacked by retrieved content, runawayloops, sandbox escapes. We are hiring the engineer who owns thisproblem.

The role

You will research, develop, deploy, operate, and analyze the safetystack for agentic AI at Janelia:

• Sandboxing. Hardened, reproducible executionenvironments for agent code (kernel- and VM-level isolation,network egress control) on our OpenShift platform and HPC cluster,validated by adversarial testing.
• Guardrails. Policy enforcement on agent inputs, outputs,and tool calls; judge models; human approval gates; independentsafeguard services that sessions can be routed through,individually or chained.
• Observability. Attributable audit logging, full-tracecapture, real-time monitoring, and post-incident analysis of agentsessions.

You will join early enough to make the foundational architecturedecisions, and you will operate what you design. We expect andsupport open-sourcing and publishing this work.

The environment

A large NVIDIA GPU cluster (B300/H200/H100 class) runningself-hosted open-weight models and frontier commercial models;collaborations with industry (Anthropic and Google); scientistsacross biology, physics, and ML as your users; the long-termstability of a philanthropy-funded research institute.

What we look for

• Hands-on experience operating LLM agents and dealing with theirfailures, or deep sandbox/container-security expertise with clearability to move into the agentic domain.
• Linux and Kubernetes/OpenShift isolation depth, including aprecise understanding of the difference between containerizationand sandboxing.
• Working knowledge of agent architectures: tool-callingprotocols (e.g., MCP), approval gates, judge models, guardrailframeworks, and the limits of each.
• Evidence over certifications: open-source contributions,publications or technical writing, disclosed findings, orproduction systems you can discuss in depth.
• Minimum 5 years in security, systems/SRE, or ML systemsengineering.

Physical Requirements

Remaining in a normal seated or standing position for extendedperiods of time; reaching and grasping by extending hand(s) orarm(s); dexterity to manipulate objects with fingers, for exampleusing a keyboard; communication skills using the spoken word;ability to see and hear within normal parameters; ability to moveabout workspace. The position requires mobility, including theability to move materials weighing up to several pounds (such as alaptop computer or tablet).

Persons with disabilities may be able to perform the essentialduties of this position with reasonable accommodation. Requests forreasonable accommodation will be evaluated on an individualbasis.

Please Note:

This job description sets forth the job’s principal duties, responsibilities, and requirements; it should not be construed as an exhaustive statement, however. Unless they begin with the word "may", the Essential Duties and Responsibilities described above are “essential functions” of the job, as defined by the Americanswith Disabilities Act.

Hiring Pay Range

Agentic AI Security Engineer: $149,084.80 - $186,356.00

Agentic AI Security Senior Engineer: $181,143.20 -$226,429.00

Agentic AI Security Principal Engineer: $210,803.20 -$263,504.00

Pay Type

Salary

The posted range reflects HHMI’s good faith estimate of the anticipated hiring salary range for this role at the time ofposting. Actual hiring compensation is determined by a candidate’s qualifications, experience, and internal equity.

Compensation and Benefits

Our employees are compensated from a total rewards perspective in many ways for their contributions to our mission, including competitive pay, exceptional health benefits, retirement plans, time off, and a range of recognition and wellness programs. Visitour Benefits atHHMI site to learn more.

HHMI is an Equal Opportunity Employer

We use E-Verify to confirm the identity and employment eligibility of all new hires.