Senior AI Engineer

About the role

This is a development role. You will design and build the AI services at the center of our platform's next phase, and you will own the technical direction for how AI works in our system. Because you will also be the deepest AI person on the team, part of the job is bringing our developers along as they start working with AI themselves.

We are embedding AI into the platform itself — assisting users with their work and supporting decision making inside automated processes. That means AI output has real consequences for real people, and the engineering standard is correspondingly high. Results have to be measurable, repeatable, and defensible, not merely impressive in a demo.

This is not a research role and it is not a prompt-writing role. It is for someone who has been building with AI daily, in real systems, and can stand behind their design decisions alongside architects who know this system well.

What You'll Do

Design and build our model routing layer. A service that accepts REST calls from our platform, routes requests to the right model, and handles communication with AWS Bedrock. You will own the design, stand behind it, and work it through with our existing architects. Language and framework are open — you help make that call based on the requirements.

Build the AI capabilities users actually touch — task assistance and decision support embedded in automated business processes, along with document analysis. Getting these right means structured and validated output, sensible fallback when a model misbehaves, and a real way to measure whether the results are correct.

Grow the AI layer over time into more sophisticated orchestration, including grounding AI responses in our operational policies through RAG or other retrieval approaches.

Contribute to development work across the program as the need arises.

Lead our adoption of AI-assisted development. Our developers are capable engineers who are new to working with AI. You help them adopt sound practice — context management, planning, AI memory, and how AI-assisted development works on a shared codebase rather than a solo one. Some of our teams work on a low-code automation platform that has just released its own MCP server and AI-assisted development features; they know the platform deeply, but not AI. You bridge that gap — platform expertise is not expected of you.

Advise on architecture for AI code assist across our stack so we set it up correctly the first time, and evaluate the tooling as it changes.

Share what you know as you go. One-on-one sessions, the occasional brown bag, helping people stand up their AI development environments. Some of this extends to our government client’s staff.

Work as a member of a scrum team alongside two partner contractors and our government client, participating in normal agile ceremonies.

AI Depth — Our Primary Requirement

We will weight this above any other qualification. We are looking for someone who has built with AI in production, not someone who has experimented with it.

You have built and shipped retrieval-augmented generation systems, agentic workflows, or applications that call models programmatically — and can walk through the architecture, what failed, and what you changed.

You have evaluated AI output systematically. Test sets, regression checks when prompts or models change, measuring whether retrieval is returning the right things. You know that “it looked good when I tried it” is not evidence.

You have engineered around non-determinism — structured output, schema validation, retries, fallback behavior, prompt versioning.

You have managed context, cost, and latency as real constraints, not abstractions.

You use agentic coding tools in real repositories every day, and have views on how a team adopts them without degrading the codebase.

You understand how models are trained and how inference produces a response, well enough to explain to a developer why a model was confidently wrong. Building one from scratch isn’t expected; knowing how one gets built is.

You have thought about what is and isn’t safe to put in a prompt, and what a model-driven process should be permitted to do on its own.

Be prepared to walk us through something you built.

Also Required

Active Top Secret clearance.

Bachelor’s degree and 10+ years of professional technology experience, including delivery of business applications.

Strong software engineering fundamentals and real depth in at least one modern language — C#, Java, Python, TypeScript or similar. Our stack includes .NET/C#, Angular, Python, and React, and you should be comfortable moving between languages. We care far more about how well you engineer than which language you engineered in.

Solid API and service design experience.

The people skills to lead as an expert — patient with developers who are new to AI, collaborative with architects who have years of experience on this system, comfortable in front of a government client.

Preferred

AWS Bedrock experience. Our available models are those offered through Bedrock.

MCP server development or integration.

React, or recent front-end work in a comparable modern framework.

Prior work in a classified or otherwise regulated environment. You’ll need to learn and work within federal AI rules on this program; prior exposure shortens the ramp.

Direct-hire position 200k+ with world-class benefits (100% paid health care for you and your family, 401k match, 4 weeks of paid leave + holidays)