Senior AI Engineer

What you'll do:

Design and ship AI solutions - agent orchestration, tool use, memory, retrieval, and the integrations that connect them to SNHU systems of record and data.
Oversee architectural decisions and their documentation (Architectural Decision Records), making choices the host team or sustainment owner can maintain after the pod departs.
Design and operate solution-specific tests and evaluations; build the instrumentation that makes offline and in-product signals actionable.
Implement agent and orchestration patterns using frameworks like LangGraph/LangChain, CrewAI, AutoGen, MCP, or custom approaches choosing pragmatically based on what the solution and the host environment can sustain.
Surface delivery and technical risk early; make scope and architecture trade-offs decisively rather than letting them drift.
Use AI development tools like Claude Code, Cursor, Codex, GitHub Copilot, and latest tools as a daily part of how you ship. Maintain a clear point of view on where they help, where they fail, and how to validate their output.
Pair with the host team's engineers on AI-augmented development practice, so they leave the engagement more capable, not just better served.
Produce the standard pod artifact set, which includes Agent Design Briefs, ADRs, Evaluation Scorecards, Deployment and Operations Runbooks, and retros. Ensure the quality bar is high enough to allow the next pod and the next domain team to reuse what you built.
Contribute to shared AI Engineering assets - prompt and skill libraries, evaluation templates, reference architectures, practice handbook entries - that codify what works and increase capabilities across the team.
Pod collaboration The canonical pod has three peers with joint accountability:

AI Analyst - owns initiative framing, requirements craft, solution assessment, and lightweight delivery facilitation. You partner with them on these and contribute your AI-feasibility judgment, but you don't own them.
Domain Product Owner - seated from the host team; carry domain knowledge and owns outcomes on the host side. You and the AI Analyst work with them on what to build and how it lands in the operational reality of the host team.
AI Engineer (you) - feasibility, build, AI architecture, evaluation. You participate in discovery and shaping from the start - this is not a bring-me-a-spec role. You contribute to sprint planning, demos, and retros at the pace the pod sets, and mentor less-senior engineers on the pod and across AI Engineering.
Partner with SNHU AI on governance, safety, evaluation, and adoption as they relate to your build work. SNHU AI owns governance; you collaborate on it and design with it in mind.
Coordinate with IT on infrastructure, IAM, security review, and DevOps as pod work requires.
Align with the host team and adjacent functions where mutual coordination makes sense, without duplicating their steady-state work.

What we're looking for:

5+ years shipping and operating production software at scale in one or more modern languages. We hire for engineering judgment and shipped systems, not a specific stack. We use Python as our working language; we don't require deep prior Python, but you'll need to get productive in it quickly.
Production experience in at least one major cloud (AWS, Azure, or GCP).
Hands-on production work with LLMs - orchestration, memory, tool use, evaluation.
Experience designing agentic systems. These systems involve tool use, state/memory, orchestration, and evaluation. Additionally, familiar with at least one framework, such as LangGraph/LangChain, CrewAI, AutoGen, MCP, or a custom approach.
Practical LLM evaluation - designing evals, reading offline vs. in-product signals, building the instrumentation to act on them.
Infrastructure as Code (CDK, Terraform, or similar) and CI/CD (GitHub Actions or similar) - self-sufficient, not a specialist.
Experience with a relational database such as MySQL, PostgreSQL, or Oracle.
Demonstrated hands-on proficiency with AI-assisted development tools, such as Claude Code, Cursor, Codex, and GitHub Copilot. These tools accelerate delivery in certain areas, but they also introduce risk or inaccuracy in others. Ability to guide and validate outputs to avoid and detect issues that can impact results.
Experience translating ambiguous domain problems into shipped systems with non-engineering partners.