Staff Full Spectrum Cyber AI Researcher

Staff Full Spectrum Cyber AI Researcher

Leidos Cyber Accelerator is actively seeking a hands-on applied researcher to invent, prototype, and operationalize agentic AI systems, alongside the automation required for their deployment in enterprise environments. This role involves evaluating emerging agentic patterns such as multi-agent coordination, tool use, and human-in-the-loop oversight. You will implement these as working systems and develop reusable automation and integration patterns, including LLM-driven tool connectors/protocols, to enable secure, observable, and repeatable paths for deployment, validation, and security testing.

Primary Responsibilities

• Be comfortable with modern tools for configuration and software development (e.g., Codex, Claude Code).
• Research, prototype, and evaluate multi-agent architectures (supervisor/worker, hierarchical, HITL) and translate successful patterns into reusable libraries/services using agent/orchestration frameworks (e.g., LangGraph-style state machines, AutoGen-style agent teams, CrewAI-style workflows).
• Build secure agent-to-tool integrations using open connectivity standards and agent tool-integration protocols (e.g., Model Context Protocol (MCP), structured tool/function calling patterns).
• Create deployment automation for agentic services, focusing on reproducible environments and repeatable enterprise releases on Kubernetes using IaC & environment provisioning (e.g., Terraform) and Kubernetes packaging & configuration management (e.g., Helm, Kustomize).
• Stand up production-like stacks on AWS + Kubernetes and codify them into repeatable, automated tooling.
• Debug and refine AI-driven workflows, ensuring work can be generalized.
• Create trials and collect metrics to demonstrate the success of automated deployments and connectivity between enterprise components.
• Build observability-first agent systems, incorporating tool-call telemetry, step tracing, and evaluation hooks.
• Research and apply nonhuman/workload identity patterns for agentic services and integrate them into deployment automation, leveraging model serving/inference platforms (e.g., Ollama, vLLM, Ray Serve) and observability/instrumentation standards (e.g., OpenTelemetry) with common backends (e.g., Prometheus, Grafana, Jaeger).
• (Optional, as mission needs) Experiment with scalable serving patterns for LLM endpoints on Kubernetes.

Basic Qualifications

• Bachelor's degree and 8+ years of relevant experience in software engineering, applied R&D, platform/DevSecOps, AI/ML engineering, or cyber engineering. Additional years of experience may substitute for a degree.
• Proven hands-on delivery: demonstrated prototypes that ran, tools others used, and automation others depended on.
• Strong development skills; experience building LLM-enabled systems and/or multi-agent workflows.
• Practical experience with Kubernetes and modern deployment/IaC practices; familiarity with Terraform and Kubernetes.

Preferred Qualifications

• Experience productizing internal platforms, including templates, reference architectures, golden paths, and guardrails.
• GitOps delivery experience (e.g., Argo CD/Flux) and observability pipeline experience (e.g., OpenTelemetry).
• Workload identity / service-to-service authentication experience (e.g., SPIFFE/SPIRE, mTLS, policy enforcement).
• Familiarity with research methodologies, with a focus on establishing validation of techniques (e.g., F1 scores, ROC/AUC, correlation methods).
• Ability to obtain and maintain a TS/SCI clearance.

Key skills/competency

• Agentic AI Systems
• Multi-Agent Architectures
• Kubernetes
• Infrastructure as Code (IaC)
• Large Language Models (LLM)
• Cyber Security
• DevSecOps
• Observability
• AWS Cloud
• Applied Research