Agentic Systems Full-Stack Software Engineer, Senior

Booz Allen Hamilton•Fayetteville, NC

5d•$86,800 - $198,000

About The Position

Most engineering roles ask you to write software. This one asks you to design it, direct it, and stand behind it. As a full-stack software engineer, you can take a problem from vision to production-ready in a fast, agile environment. If you're looking for the chance to not just develop software, but to help design and create systems that make a difference, we need you on our team. You'll work in an agentic-first codebase, where a large share of the code is drafted by AI agents working from a specification you write, then reviewed, tested, and shipped by you. The job is less about typing and more about judgment, decomposing work into verifiable pieces, writing the spec and the tests that constrain the agent, and reviewing what comes back with the same rigor you'd apply to a teammate's work. Engineers who can only prompt struggle here. So do engineers who insist on handwriting everything. We're looking for the one who does both well, and who can tell the difference between output that looks right and output that is right. You'll help clients overcome their most difficult challenges using the latest architectural approaches, tools, and technologies. You'll make sure the solutions your team delivers account for the current architecture and operating environment as well as future functionality and enhancements. And you'll help define how software actually gets built over the next five years, rather than just adopting someone else's workflow.

Requirements

5+ years of experience with full-stack development, including a Python web framework such as FastAPI, Flask, or Django and a modern JavaScript framework such as React, Vue, or Svelte
3+ years of experience with cloud computing, including AWS, GCP, or Azure, and relational databases, including schema design, query performance, and versioned migrations
2+ years of experience with Kubernetes
1+ years of experience with agentic engineering using Claude Code, Codex, Cursor, or tools including writing project-level agent instructions, decomposing work into agent-executable specifications, and reviewing AI-generated code before merge
Experience with Docker, Helm, and infrastructure as code, including Terraform, and managing and deploying LLMs in a Kubernetes environment
Experience with LLM application patterns, including retrieval-augmented generation, vector databases, tool or function calling, or streaming inference
Experience with Agile development and Git-based collaboration workflows, including issue tracking, merge requests, code review, and CI/CD pipelines such as GitLab or GitHub
Ability to obtain a TS/SCI clearance
Bachelor's degree
Ability to obtain a DoD 8570 IAT Level II Certification within 2 months of hire date

Nice To Haves

Experience managing and deploying applications in an air-gapped environment
Experience with attribute or policy-based authorization systems, including AWS IAM, Cedar, OPA, Rego, or Casbin
Experience with data handling in classified environments, including classification markings, compartments, or clearance-based access control
Experience with GPU workload scheduling on Kubernetes, including time-slicing, MIG, or device plugins, OpenTelemetry, and distributed tracing across a microservice fleet
Experience building evaluations, benchmarks, or regression harnesses for LLM-backed features
Experience creating and conducting automated unit and system integration tests to identify and provide feedback on failed modules and components
Experience contributing to open-source developer tooling, agent frameworks, or MCP servers
Possession of excellent verbal and written communication skills, to communicate technical ideas and deliver briefings with cross-functional teams
Master's degree
Security+ Certification

Responsibilities

Ship features end-to-end across ten Python and FastAPI microservices and a React or TypeScript front end including database migration, service logic, API contract, UI, and tests in a single change.
Author and maintain the agent harness that makes this codebase productive for AI agents including project-level instructions, reusable skills, specialized subagents, and the evaluations that tell you when the harness has regressed.
Extend a multi-tenant authorization system modeled on AWS IAM including policy evaluation order, organizational guardrails, cross-organization consent, per-service database isolation, and audit trails that hold up under inspection.
Deploy and operate LLM inference on Kubernetes including vLLM on GPU nodes, Helm charts, model routing, and per-request usage and cost accounting.
Own infrastructure as code including Terraform modules for databases, caches, object storage, queues, and secrets across multiple environments.
Promote releases through a staging → production → air-gapped enclave ladder, including packaging for environments with no internet egress.
Review AI-generated changes for correctness, security, and architectural fit and when a class of mistake recurs, fix it in the harness rather than in the diff.