Principal Engineer - Agentic AI

About IRIS

IRIS Software Group is one of the UK's largest privately held software companies, trusted by 100,000+ businesses, schools and accountancy firms to keep their operations running. Our software pays 1 in 6 UK employees, supports over 12,000 schools, and is relied on by 91 of the top 100 UK accountancy firms.

We're a Great Place to Work® certified employer, recognised for our commitment to well-being, inclusion and development - and we're growing fast.

IRIS is seeking a Principal Engineer to play a pivotal role in accelerating the adoption of AI-driven, agentic software development across its engineering organisation. As part of the Engineering Transformation Programme, this individual contributor role combines deep technical expertise with hands-on enablement, working directly alongside engineering teams to embed new AI-powered development practices into day-to-day delivery.

Rather than focusing on training alone, the role bridges the gap between learning and adoption. The successful candidate will work across multiple product divisions, helping teams overcome real-world challenges, while shaping best practices, influencing programme strategy, and driving measurable improvements in engineering productivity. This is a highly visible position, partnering closely with senior leaders across Engineering, Agility, and AI transformation, and offers the opportunity to help define the future of software development at IRIS.

Core responsibilities:

• Embed directly with pod teams across divisions following bootcamp completion, working alongside engineers on real delivery work and enabling adoption of agentic tooling and workflows.
  This is hands-on engineering, not observation.

• Diagnose and debug the specific blockers teams encounter with agentic adoption - whether that is tooling configuration, context engineering, workflow integration, codebase-specific challenges or confidence and mindset. Solve problems with teams, not for them.

• Build and maintain a living body of best practice: what works, what does not, what is codebase or context-specific versus universally applicable. Actively share this across the estate - through the engineering community, cascades to SLT and divisional leads.

• Partner with the Director of Enterprise Agility and the Agile Coach team to ensure the technical and human change elements of adoption are joined up. Where an adoption challenge is partly a ways-of-working problem and partly a tooling problem, work across both.

• Partner with the Director of Agentic AI on training content, tooling setup and the ongoing evolution of the agentic programme - providing the ground-level feedback loop that keeps programme design connected to what actually benefits the teams.

• Support the configuration and integration of agentic tooling within divisional environments - including MCP integration, IDE setup and workflow automation - to the point where teams can self-serve and maintain it.

• Evaluate agentic tooling options - Cursor, Claude Code, GitHub Copilot and emerging alternatives - against real adoption data and hands-on usage from embedded assignments to help shape the programme's tooling decisions with evidence-based recommendations.

• Create, maintain and share a library of common agentic skills and workflows - reusable prompts, skill definitions and automation patterns - so that teams across divisions can adopt proven approaches directly rather than rebuilding them from scratch.

• Act as the go-to expert across the engineering estate for agentic engineering questions. Be genuinely reachable and collaborative.

• Operate across multiple divisional contexts over time, moving between assignments as adoption matures in one area and the need shifts to another. Comfortable switching context, picking up unfamiliar codebases, and building credibility quickly with new teams.

What Success Looks Like:

• Sustainable team adoption of agentic development practices, with bootcamp graduates using agentic tooling as a natural part of their daily workflow rather than as an occasional experiment. Embedded assignments have made a measurable impact on both the speed and depth of adoption.

• A documented and actively maintained best practice library, providing practical guidance, reusable patterns, and proven approaches drawn from real-world experience across multiple divisions, teams, and codebases.

• A strong feedback loop between delivery teams and programme leadership, enabling more informed decisions on tooling, training, rollout sequencing, and investment priorities based on real adoption outcomes.

• Increased capability across engineering and agile communities, with Engineering Leads, Agile Coaches, and other stakeholders better equipped to support and sustain agentic ways of working within their teams.

• Clear visibility of adoption maturity across the organisation, including where agentic practices are thriving, where additional support is required, and which interventions will deliver the greatest value in the next phase of the programme.

Future Growth & Impact

• Helping shape and optimise advanced agentic engineering practices, including areas such as multi-agent workflows, automated testing, and agentic code review.

• Identifying and influencing future capability-building and training requirements as teams continue to mature beyond the initial bootcamp phase.

• Contributing to the development of a broader engineering enablement function, supporting the evolution from agentic adoption into wider developer experience (DevEx) and engineering productivity initiatives across the estate.

Experience

• Deep, hands-on experience with agentic development tooling in a professional engineering context — Cursor, GitHub Copilot, Claude Code, or equivalent.
  Not just familiarity, but sustained daily use on real codebases, with a clear point of view on what works and where the limits are.

• Strong software engineering background across at least one primary language and stack. Sufficient breadth to pick up an unfamiliar codebase and be productive within it quickly.
  The role requires genuine engineering credibility, not just knowledge of the tools.

• Experience working across multiple teams or contexts simultaneously, or moving between assignments in a consultative or enabling capacity.
  Comfortable being the new person in a team repeatedly and building trust and credibility quickly.

• A track record of solving real engineering problems - not designing solutions in the abstract, but sitting with a team, understanding their specific situation and working through it with them.

• Familiarity with modern software delivery practices: CI/CD, trunk-based development, automated testing, observability tooling.
  Enough context to understand how agentic workflows fit into - and sometimes challenge - existing engineering norms.

Approach

• Does the work first, explains it second. Earns credibility with engineers by being genuinely useful on real problems, not by presenting frameworks or running workshops.

• Meets teams where they are. Understands that adoption challenges are rarely just technical - they are also about confidence, habit and context - and adjusts accordingly without losing focus on the engineering substance.

• Curious and honest about what agentic tooling can and cannot do. Does not oversell, does not hedge. Has a clear, experience-based view of where AI-assisted development genuinely accelerates and where human judgement remains the right call.

• Shares knowledge actively and generously. Writes things up, contributes to the community, flags patterns to the Director of Agentic AI - because individual impact at team level compounds when it is shared across the estate.

• Comfortable with context-switching and ambiguity. Each embedded assignment brings a different codebase, a different team culture and a different set of adoption challenges. Treats that variety as interesting rather than disruptive.

• Brings a solution-based mindset and helps teams develop the same. Moves the conversation from “this doesn’t work for us” to “let’s figure out why and how to fix it”.

Useful Experience

• Experience with MCP (Model Context Protocol) server configuration, tool integration or agentic workflow automation beyond IDE-level tooling.

• Exposure to prompt engineering, agent orchestration or multi-agent system design at a practical level.

• Experience in a B2B SaaS or multi-product software organisation, particularly one with a mix of modern and legacy codebases.

• Familiarity with the team topologies model and how enabling functions operate within a distributed engineering organisation.

• Experience in regulated or compliance-sensitive software environments (accountancy, payroll, HCM, or education).

Why You’ll Love Working Here

• Impact: Your work will influence millions globally.

• Growth: Continuous learning and career development opportunities.

• Belonging: A culture that celebrates diversity and empowers every individual.
  

Ready to Apply?

Click Apply - we’re excited to learn about your unique perspective and experience. If you need adjustments during the process, let us know. We’re committed to making this opportunity accessible to everyone.

What to expect from our hiring process

Our process is designed to be fair, transparent and straightforward. Stages vary depending on the role — more senior positions may involve additional steps, and some areas include role-specific assessments such as a technical test or case study — but typically you can expect:

1. Application review — we assess your experience and potential

2. Initial call — a short conversation to learn about you and share more about the role

3. Skills assessment — tailored to the position, e.g. case study, coding challenge or portfolio review

4. Final interview — meet the team and explore how you'll make an IMPACT at IRIS

We'll always walk you through the specific stages at the start of the process so you can prepare with confidence.

If you need any adjustments or accommodations during the process, let us know, we’re committed to making this experience accessible for everyone.