AI Project Manager

Vacancy No VN603 Basis Permanent Full Time / Part Time Full Time Entity Aventum Group Management Services Ltd Location City London Job Details Role Summary We’re looking for an AI Project Manager to orchestrate delivery across our AtomX product portfolio a family of insurance products built in parallel by multiple AI-native engineering teams, each led by a Tech Lead and paired with a Product Manager, under a common architecture. We work in a spec-driven development model: business intent is captured as a precise, signed specification and an agreed test pack before any code is written, with AI agents handling much of the implementation and testing. That model has changed the economics of delivery building is no longer the slow part. What can slow us down now is the space between the products: shared reference data, integrations, common components, migration sequencing, and capabilities that depend on each other across teams. Your job is to own that space. The primary responsibility of this role is to proactively understand the interdependencies among the products and get them resolved before they surface as blockers. That requires someone technical enough to read an integration spec, follow a data flow, and hold a credible conversation with Tech Leads and Architecture, alongside the full range of classic project management disciplines: planning, cadence, risk, dependency and stakeholder management. This is a coordination and delivery role, not a line-management or people-management role the Product Managers own their products, and the Tech Leads own their builds. You own the picture across all of them. This is an AI-native organisation, and the strongest signal we can get is not a certification it is what you have already built to make yourself more productive. We will ask about this at interview, and we would expect to see several of the following, or projects like them: You do not need to have done all of these. What we are looking for is the instinct: when you meet repetitive work, your first move is to remove it and you use AI and automation as naturally in your own work as we use them in engineering. Role Accountabilities * Delivery planning. Own the cross-portfolio delivery plan: the sequencing of capabilities from specification sign-off through build, UAT and release, across all product teams and keep it honest as things move. * Dependency mapping. Proactively map the interdependencies among the products shared reference data and lists, integrations and interface versions, common components, data flows between products, and capabilities whose builds depend on one another and maintain that map as a living view the whole team trusts. * Dependency resolution. Get dependencies resolved, not just recorded: broker the agreement between the Product Managers and Tech Leads involved, drive shared facts to a single settled answer, and escalate through the agreed route when ownership is disputed before the dependency reaches a build. * Operating rhythm. Run the delivery cadence across the product teams: progress against signed specifications, open items lists with names and dates, blockers on both sides made visible, and work-in-progress limits respected so the same people are not consumed by every workstream at once. * Shared resources. Coordinate the people and capabilities shared across teams front-end engineers, shared AI engineers, platform and DevOps, and Architecture’s involvement in integration, migration and reference-data decisions so no team stalls waiting for a shared resource nobody scheduled. * Third parties. Track third-party integrations across the portfolio: specs signed by both sides against the same version, test environments in place, and failure and reconciliation behaviour agreed flagging early where an external party is the critical path. * Migration and release. Plan and coordinate migration and cutover sequencing with Architecture and the business, including parallel runs and reconciliation checkpoints, across products that share data or processes. * Guardrails. Protect the operating model: nothing enters build until its specification is closed and its test pack agreed, and changes after sign-off go through the change route rather than becoming silent mid-build corrections. Pace is the point discipline is what keeps it safe. * Risk. Maintain a risk and dependency log built for AI-native pace, where feedback cycles run in days and hours rather than months and where a risk raised late is a risk realised. * Reporting. Provide clear, honest reporting to the Chief AI Officer, business sponsors and management teams: per capability, per product, with open items and their build effect stated rather than smoothed over. * Automated your own reporting. A script, agent or workflow that assembles your progress reports, RAID logs or steering packs directly from your tools (ADO, Jira, or similar) rather than compiling them by hand each week. * Built a live view of your portfolio. A live dependency map, delivery dashboard or risk view driven from tool APIs or exports something that stays current on its own, rather than a slide deck you maintain manually. * Made meeting output automatic. Used AI tools (e.g. Claude, ChatGPT, Copilot) to turn meetings and threads into structured minutes, actions, decisions and follow-ups — as a working routine you rely on, not a one-off experiment. * Automated the chasing. Bots, reminders or small workflows (Power Automate, Zapier, scripts) that follow up open items, chase approaching dates and flag ageing blockers, so that nothing depends on you remembering. * Prepared for technical conversations with AI. Used AI to read and summarise technical artefacts API specifications, schema changes, release notes, test evidence to arrive at engineering conversations already informed. * Built yourself an assistant with context. Set up an AI assistant, project or agent loaded with your programme’s context plans, terminology, standards that drafts plans, comms and risk assessments in your voice for you to refine rather than write from scratch. * Removed a manual reconciliation. Written or AI-generated small scripts (Python, PowerShell, or similar) to reconcile data between two systems, merge trackers, or produce a report nobody wanted to compile manually. Role Requirements * Proven experience as a AI Project Manager, Delivery Manager or equivalent in software delivery, ideally within insurance or financial services. * Demonstrable track record of managing dependencies across a multi-team product portfolio not a single-team delivery and of resolving cross-team contention rather than reporting it. * Technical enough to work at the engineering boundary: able to read an API or integration specification, follow a data model and a data flow between systems, understand environments, releases and CI/CD, and hold a credible technical conversation with engineers and architects about integration design, failure behaviour and migration. You will not be asked to code, but you must be able to follow the build. * Experience with specification-gated or milestone-gated delivery alongside agile methods, and the judgement to know which discipline a situation needs. * Comfort working in AI-assisted delivery reviewing agent-generated outputs and test evidence, and working at a pace where a build cycle is measured in days. * Strong stakeholder management up to executive and sponsor level, including brokering agreement between parties who want different things. * Exceptional written precision: a dependency, risk or decision recorded ambiguously is one that will be argued about later. * Familiarity with tools like ADO, Jira, Confluence and analytics platforms. * Bonus: exposure to MGA or Lloyd’s market operations; experience of legacy-platform migration programmes; or experience coordinating shared platform, RPA or automation teams. Management Duties * No We are an equal opportunity employer, and we are proud to share that 93% of our employees say they can be themselves at work. We aim to hire our industry's finest people because the best people drive the best outcomes. And we forever challenge the status quo because we know there are always ways to improve things. Because together, we're limitless. We value applicants from all backgrounds and foster a culture of inclusivity. We understand the need for flexibility, so work in a hybrid model. Please let us know if you require any reasonable adjustments during the recruitment process.