IT Operations Lead - Banking

EggAI · Banking transformation programme (UK) · Remote, UK-facing — London for gate weeks

About the Programme

EggAI is the transformation partner on a core banking system replacement for a client to lay the foundations of operating Agentic Generative AI in production.

EggAI's role is to specify, set the standard and programme-manage. You will define what operational readiness means and prove it at each gate, working with the client's own teams and its managed service providers, as well as EggAI's platform engineers.

The platform is a mix of classical banking infrastructure and agentic AI workloads, so the reliability problem runs across both deterministic and probabilistic systems.

The Role

Own the definition of operational readiness for go-live: what "ready to run in production" means, how the bank proves it at each regulatory gate, and what has to change in the organisation for that proof to hold.

The bank's operations teams, processes and provider contracts were built to run classical banking infrastructure. Agentic AI workloads need a different operating model, and you will shape it: redraw responsibilities between the client's teams and its managed service providers, set the standards each party works to, and lead the people through the change — so that those who run the platform today are ready and willing to run what comes next.

This is a lead/principal position reporting to the Programme Plan and Execution Lead. It is remote, with regular London presence required for gate weeks and workshops.

Responsibilities

Reliability targets — set the SLO and error-budget framework alongside the client's existing structure, have the service owners adopt it, and commission the behavioural quality measures that do the same job for agentic workloads

Observability and detection — define what Tier-1 services and agentic workloads must emit, including trajectory logging; require an automated path from a firing alert to the affected services and their owners; and hold the operating teams to closing the logging gaps in what runs today

The incident and change path — set the runbook standard and its required coverage, extend incident classification and intervention mechanisms to agentic failure modes, and have rollback and change-failure rates measured and reported by those who own them

Capacity and performance — direct the headroom position through the parallel-run peak at cutover, sponsor the chaos and game-day programme for payment rails and the core ledger, and mandate that capacity modelling covers inference workloads

Gate evidence — own the technical content of the Operational Acceptance Report and Operational Readiness Certificate, the reliability section of each gate pack, and the reliability NFRs defended as ADRs through the client's Architecture Design Authority — written by the teams, reviewed and signed off by you

The operating organisation — decide how responsibilities sit between the client's teams and its managed service providers for the new platform, and lead the people concerned through that change

Expected experience

10+ years in leading IT operations teams — You have operated L1, L2, and L3 support as SRE, platform reliability or infrastructure organisations, with the people and the providers reporting to you

Aligns operations with development, engineering and the business — has set operating goals that engineering teams and business stakeholders signed up to, and kept them focused on those goals through delivery

Has introduced SLOs and error budgets into an organisation that did not have them, and can build on the experienced success factors

Infrastructure in depth — has deep infrastructure knowledge, and what is required for smooth business operations: telemetry, redundancies, backups, etc.

Incident management, with post-incident reviews that led to real changes in the system and in the organisation

Regulated financial services, or another environment where an external body audits operational evidence.

Has worked through a managed service provider and can set standards they do not personally execute

Comfortable operating in a complex, multi-stakeholder programme where scope and ownership are still being established across parallel workstreams

Valuable, not required

LLM or agentic system operations: eval harnesses, drift detection, trajectory analysis, prompt and model versioning treated as change types

Core banking or payments exposure, particularly scheme gateway integration

What This Role Is Not

Not a platform architect. A separate role owns agentic platform design and build

Not an on-call engineer. You will not carry a client pager; out-of-hours operations are run by a third party

Not a service delivery manager. This is engineering judgement applied to readiness, not ITSM process administration