Product Owner

You will own the behaviour of a specific AI system or workflow inside a larger product. You will work with a team and a defined set of internal stakeholders. Your primary stakeholders are business teams who will use the workflow and engineering teams who will build it. You will be responsible for creating a spec that the business teams can approve and engineering teams can build efficiently. Your success is measured by whether the AI system behaves as specified in production over time.

You will work with QA and engineering to build and review evaluations. This is a critical part of the role. Evaluations are how you verify the AI system keeps behaving as specified once it's live.

Building a working understanding of how AI systems work, and what that means for product and analytics, is critical to this role. You will also act as a proxy for the business team and run UAT.

Alongside that, the role carries the usual weight of product ownership: picking up an unfamiliar domain fast, holding the line on scope, and keeping stakeholders aligned on what is shipping and when.

What you will own

• Own and prioritise the backlog for your system; run refinement, planning, and acceptance with the delivery team. You could also be running kanban boards for your team for a fast paced project.
• Turn business requirements into AI system requirements for the engineering team.
• The requirements should meet the Jeavio guardrails of AI governance, privacy, security and responsible AI standards, and highlight it when a requirement falls outside them.
• Support adoption: training, workflow changes, review Evals, user manuals, release notes etc. that might be required to get the feature to production.

PO Competency

• Pick up a new domain quickly and get to the point where you can ask informed questions about the business.
• Should be able to adopt AI tools like Claude or similar to work upon eliciting and/or prototyping requirements (including UX designs) independently or collaboratively with technical peers.
• Should be able to participate in and lead (where appropriate) discovery sessions to be able to understand the lay of the land, vision and translate it back into documentation that can be used by both the business and engineering teams.
• You should be able to figure out which are the relevant metrics for users and business.
• AI can be more accurate but slower, more flexible but less predictable, cheaper but occasionally wrong. You should be able to weigh tradeoffs from a product point of view - i.e. whether a certain trade off is okay or breaks user experience.
• The Evals you define must align with the product requirements.

Stakeholder Management

• Steer stakeholders towards the better option or push back on scope during scope based negotiations by working alongside the engineering team.
• Be able to influence the roadmap by understanding what the team can deliver vs what the stakeholders want.
• Should also be able to present project status to stakeholders and give an update on delivery timelines by aligning with the technical teams.
• Should be able to understand and communicate the limitations of the system, clearly to stakeholders.

Behavioural specification

• Write acceptance criteria for probabilistic outputs.
• Work with Engineering and QA colleagues to define system guardrails.
• Specify the confidence thresholds that trigger human review, a fallback path, or a graceful exit.
• Read and assess system prompts well enough to tell whether they match the spec.
• Write the spec for two readers at once: the AI engineer who implements it and the business user who will live with it.

Evaluation and human oversight

• Own the evaluation dataset for your system across its lifecycle.
• Interpret results and translate them for non technical stakeholders.

Security & Compliance:

• Define and prioritize security requirements in product backlog.
• Ensure data protection, privacy, and compliance with ISO 27001 policies.
• Collaborate with engineering and security teams for secure product delivery.
• Manage risks related to features, integrations, and data handling.
• Support audit readiness and continuous security improvements.

What we are looking for

Required

• 6+ years in product ownership, product management, or business analysis, including at least two years owning a backlog with a delivery team.
• Direct experience shipping or operating an AI/ML or LLM-based feature in production — not only prototyping.
• Fluency in AI tools - ChatGPT or Claude including an understanding of skills, plugins, and context management.
• Demonstrable understanding of how LLMs work and how they impact product decisions.
• Demonstrated ability to write acceptance criteria for a non-deterministic output.
• Working fluency in Evals.
• Judgment about where human in the loop use case fits in.
• Strong written communication.
• Should be able to keep up with evolving trends and is open to unlearn and learn quickly.

Good to have

• Experience with agentic systems: tool use, multi-step orchestration, retrieval, or observability tooling.
• Familiarity with our stack: AWS, Claude, LangSmith, JIRA.