> Markdown version of [/jobs/ext/3104883-software-engineer-4-ai-native](https://www.wearedevelopers.com/jobs/ext/3104883-software-engineer-4-ai-native). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md). --- # Software Engineer 4 - AI Native - **Company:** Granicus, LLC - **Location:** Saint Paul, MN, United States (Remote available) - **Salary:** $88,000.0 - $180,000.0 - **Contract:** Permanent contract - **Skills:** Artificial Intelligence, Automation of Tests, Code Review, Data Structures, Web Content Accessibility Guidelines, Production Code - **Published:** September 27, 2026 - **Apply:** https://jobs.localjobnetwork.com/apply/add/88462310/1 ## About the Role * Strong engineering fundamentals. Data structures, systems design, and testing, with the ability to read unfamiliar code quickly and assess it accurately. Agents amplify engineering judgment; they do not substitute for it. * A record of shipping production software you owned the quality of, including responsibility for diagnosis and remediation when it failed. * Hands-on experience directing coding agents on production work, including their failure modes and the practice of reviewing agent-generated code critically rather than approving it by default. * Demonstrated code-review competence. You identify defects that automated tests do not catch, provide actionable feedback, and maintain a high bar without becoming a bottleneck. * High autonomy. You advance work without step-by-step direction and escalate issues proactively. Preferred * High-assurance or regulated experience. Shipping within FedRAMP, defense, financial services, healthcare, or another NIST 800-53 / SOC 2 / HIPAA-bound environment. * Depth in evaluation authoring or test-first development within a rigorous engineering culture. * Full-stack range, sufficient to review front-end and back-end agent output with equal confidence. * Public-sector or govtech experience and familiarity with the relevant end users. Indicators of a strong fit * You deliver more effectively by directing multiple agents than by authoring code directly, while retaining a precise understanding of the intended implementation. * You regard rigorous code review as a core engineering competency. * You prefer to establish quality through evaluation suites rather than through discussion. * You treat the agent array as leverage and reliably identify when its output is incorrect. * You want to contribute to how engineering is practiced here, not only to apply new tooling. ## Description When agents produce the majority of initial implementation, the scarce engineering skill shifts from authoring code to directing it, reviewing it rigorously at volume, and owning whether it is correct. This role is for a senior engineer who has made that transition: decomposing work for the agent array, authoring the evaluation suites that hold agent output to a measurable standard, reviewing agent-generated pull requests with greater rigor than most engineers apply to their own work, and shipping production code within a FedRAMP-authorized environment. The role contributes to defining how engineers operate in an AI-native lifecycle, not merely to adopting AI tooling. What Your Impact Will Look Like Direct the agent array on production workstreams - decompose problems into tasks suitable for agent execution, dispatch them, and integrate the output into shipped software. * Review agent-generated pull requests at volume and at depth - identify correctness, security, and accessibility defects that automated tests do not catch, while maintaining review throughput and a consistent quality bar. * Author evaluation suites that make quality measurable - define criteria under which the pipeline validates correctness rather than relying on subjective assessment. Eval-driven development is your standard practice. * Own quality end to end - correctness, performance, security posture, and WCAG accessibility of the software your team ships, irrespective of which component or agent produced the initial implementation. * Advance workstreams along the autonomy ladder on the basis of evidence - move work from supervised to autonomous execution when measured reliability supports it, and revert promptly when it does not. * Strengthen the development lifecycle itself - identify where patterns, prompts, or pipeline components degrade at volume and work with the lead architect to remediate them. * Maintain agent operations within the security boundary - branch-only execution, vaulted credentials, sandboxed actions, and in-VPC inference. Throughput does not justify exceptions. * OWNERSHIP FROM DAY ONE + An active workstream directed through the agent array + Agent pull-request review as a core, high-signal responsibility + Evaluation suites authored and owned for your team's deliverables + End-to-end quality ownership within the FedRAMP-authorized environment * * SCOPE YOU WILL GROW INTO + Higher autonomy ratios substantiated by reliability data + Reusable prompt and evaluation patterns adopted by other engineers + Contribution to the lifecycle standards as a co-author, not only a user + Mentorship of engineers transitioning to agent-directed delivery ## Related Videos - [Are Code Reviews Worth It? Insights from 16 Years of Review Data](https://www.wearedevelopers.com/videos/1135-are-code-reviews-worth-it-insights-from-16-years-of-review-data) - [Phel, a native Lisp for PHP](https://www.wearedevelopers.com/videos/791-phel-a-native-lisp-for-php) - [Designing the Future of Human<>Agent Collaboration](https://www.wearedevelopers.com/videos/1447-designing-the-future-of-human-agent-collaboration) - [Hiring AI Native Talents](https://www.wearedevelopers.com/videos/100268-hiring-ai-native-talents) - [The AI Velocity Trap: Shipping Faster Without Breaking More](https://www.wearedevelopers.com/videos/100326-the-ai-velocity-trap-shipping-faster-without-breaking-more) - [Dirty Tests And How To Clean Them](https://www.wearedevelopers.com/videos/515-dirty-tests-and-how-to-clean-them) ## Related Articles - [What is Agentic Programming and Why Should Developers Care?](https://www.wearedevelopers.com/magazine/625-what-is-agentic-programming-and-why-should-developers-care) - [Never delegate the understanding](https://www.wearedevelopers.com/magazine/749-never-delegate-the-understanding) - [ I Gave a Video Editor More Autonomy Than a Trading Bot. On Purpose.](https://www.wearedevelopers.com/magazine/773-i-gave-a-video-editor-more-autonomy-than-a-trading-bot-on-purpose) - [The Overflow: AI and Agentic Coding](https://www.wearedevelopers.com/magazine/721-the-overflow-ai-and-agentic-coding) - [From Prototype to Production: Build AI Agents with This Free 4-Course Learning Path](https://www.wearedevelopers.com/magazine/655-from-prototype-to-production-build-ai-agents-with-this-free-4-course-learning-path) - [What is Software Engineering in the Age of AI?](https://www.wearedevelopers.com/magazine/640-what-is-software-engineering-in-the-age-of-ai)