Blue Yonder; JDA​/RPG Consultant

Cap Nexus is an AWS consulting partner specializing in AI and cloud engineering. We build and deliver software through agentic engineering: we build and operate the systems that produce code, direct AI tooling to accelerate delivery, and exercise the judgment that keeps AI-generated work correct. Everyone at Cap Nexus works fluently inside our delivery system, which combines agentic tooling, spec-driven development, evidence-based quality, and AI-augmented delivery governance.

Cap Nexus is looking for ahands-on Blue Yonder (JDA) / RPG Consultant to support a major retail client. This is a technical consulting role focused onBlue Yonder/JDA MMS, IBM i, RPG development, DB2 for i, integrations, troubleshooting, and system enhancements.

We'relooking for someone who can work directly within the MMS environment and independently deliver technical solutions, rather than primarily functionalor end-user support.

• Build and maintain IBM i applications in free-format RPGLE, SQLRPGLE, and CL, using ILE concepts such as procedures, service programs, and modules.
• Work with DB2 for i through SQL and DDS, including physical and logical files and subfiles.
• Integrate IBM i applications with other systems through web services, JSON and XML, FTP, and the Integrated File System (IFS).
• Support changes through testing, deployment, and post-deployment, and resolve technical issues directly with client teams.
• Set coding and testing standards for IBM i work, and identify where modernizing an application pays off.
• Own the technical outcome for a system area or major feature set: propose the approach, define the task breakdown, and deliver with minimal escalation.
• Design APIs and data models that account for cost, scale, and failure, and build in observability (logging, tracing, and metrics) from the start.
• Work with Cloud Engineers on infrastructure as code and CI/CD so what you build deploys the same way every time.
• Partner with the Solutions Architect on solution design and with the Engagement Manager on delivery, and turn designs into a build plan the team can execute.
• Set code-quality standards for your area, and mentor Software Engineers I and II through pairing and reviews that teach, including how to direct AI tooling and check what it produces.
• Join architecture presentations and technical deep-dives alongside the Solutions Architect when the client needs your depth.

You own outcomes on complex work, exercise architectural judgment, and raise the bar for those around you.

• Agentic tooling. Set the tooling approach for a codebase or engagement: which work goes to a conversion engine, which is spec-driven build, where the human-in-the-loop gates belong, and how the toolchain is wired for the problem at hand.
• Spec validation and readiness. Own the quality of the handoff from specification to build. Make sure specs are executable and actually satisfy the requirements, and hold delivered work to a defined readiness bar. When generated code is wrong, fix the spec, not just the output.
• Validating AI output. Be the final judgment on correctness for high-risk changes, including in unfamiliar domains, by designing the evidence that proves behavior is preserved.
• Delivery-system ownership. Design and maintain the project-specific setup of our delivery tooling (steering, agent roles, templates, quality gates, and metrics) as durable capability, not one-off scaffolding.

Judgment over correctness:you can tell when confident, plausible-looking AI output is quietly wrong. Tooling does the mechanical work; you're accountable for whether it's right.

• 5+ years of professional software engineering experience.
• A strong grasp of architecture concepts: event-driven and service-based design, API design, data modeling, separation of concerns, and designing for change.
• Experience owning a significant system from design through production, and mentoring the engineers who built it with you.
• Working knowledge of CI/CD and infrastructure as code.
• Hands-on experience with spec-driven or agentic development workflows, where specs drive code generation, for example Kiro, Claude Code, Amazon Q Developer, or Cursor with rules. Using AI only for autocomplete or snippets doesn't count.
• A track record of catching what AI-generated or refactored code gets quietly wrong, and of building the tests and gates that prove behavior is preserved.
• Hands-on experience programming on IBM i (AS/400, iSeries):…