Applied AI Engineer

Who Are We? 
Groundswell is a premier technology integrator and solution provider, resolutely committed to solving the most complex challenges facing federal agencies today. Our name, Groundswell, represents our commitment to be an unstoppable, seismic change in government. Ours is a small company culture with big company reach and results.  Are you ready to be audacious, be bold and drive change at a rapid pace?  Join us, where we’ll make a greater impact together.

What You’ll do:

We are looking for an engineer who can take a business problem, define both the requirements and the technical approach, and deliver the result, including recognizing when AI is not the right solution.

This is a hands-on role with substantial autonomy. You will own capabilities end to end, covering the discovery conversation, the design, the build, the evidence that the capability performs, and its operation after release. You’ll work ahead of the direction you are given rather than waiting for it.

The work spans client delivery, internal product development, rapid proofs of concept, and internal enablement. You should be comfortable moving between them, and comfortable being the most AI-literate person in a room that includes engineers, executives, and end users.

This is an emerging leadership role. You will be the person engineers consult when they are new to AI, and the person clients ask about what is coming next. Maintaining current knowledge of the field is part of the role rather than something done on your own time.

Responsibilities

• Lead requirements conversations with business and technical stakeholders, covering the workflow, the decision being supported, the current standard for acceptable results, and the constraints that were not raised initially.

• Define the technical approach and defend it. Select the AI pattern appropriate to the problem, such as extraction, classification, summarization, retrieval, or an agentic workflow, rather than defaulting to the most sophisticated option available.

• Recommend against AI when a simpler solution is the right one. Rules, process changes, and improved interfaces are often the correct answer, and identifying that early is part of the job.

• Define measurable success criteria before building, including accuracy targets, human review thresholds, acceptance conditions, and the definition of failure.

• Build the complete capability rather than the AI components alone. This includes prompt and retrieval design, structured outputs, tool and function definitions, API integration, data handling, error states, and the user interface. Adoption usually depends on these supporting elements as much as on model performance.

• Build evaluation sets from real data and measure against them. Iterate based on results rather than intuition, and determine when a capability is ready for deployment.

• Make and defend architecture decisions within your scope, weighing quality, cost, latency, security, and authorization constraints.

• Operationalize capabilities in the client environment, including governance, logging, and traceability requirements. Monitor quality, cost, latency, and drift after launch, and optimize as better or less expensive options become available.

• Help clients understand what is possible. Anticipate needs, shape the next phase of work, and build the proof of concept that supports the case.

• Mentor engineers who are new to AI through code review, pairing, and guidance toward the appropriate pattern for a given problem.

• Maintain current knowledge of models, tooling, and techniques, and bring back what proves useful as reusable patterns for the team.

• Move between client delivery, internal product work, rapid proofs of concept, and internal enablement as the work requires.

• Produce clear documentation covering what the solution does, its known limitations, how it was validated, and what happens when it produces an incorrect result.

Required qualifications

• 4+ years building and shipping production software.

• At least 1 year of hands-on experience building applications that integrate large language models, including prompt engineering, retrieval-augmented generation, structured extraction, tool use, and agent patterns.

• Demonstrated ability to own a capability end to end, from an ambiguous requirement through to something in production that people rely on.

• Demonstrated experience evaluating AI system quality, including build…