• Take ideas from prototype to production. Start from a rough concept or

business problem, prove it quickly with a prototype, then do the hard

engineering to turn it into a reliable, scalable, production-grade system

people can depend on — and know what each stage demands.

• Build agentic AI, end to end. Design, implement, deploy, and evolve

autonomous and tool-using AI systems — agents that reason, plan, call

tools, and act on the user's behalf — for both internal and

customer-facing use, owning the full lifecycle and the outcome.

• Design agentic architectures. Build agent-orchestration systems: single

and multi-agent setups, tool and function calling, Model Context Protocol

(MCP) and agent-to-agent coordination, planning and reasoning loops,

memory, identity and access for agents, and the control and safety

mechanisms that keep autonomous behaviour trustworthy.

• Match the solution to the problem. No one-size-fits-all. You assess each

problem on its own terms and choose the right approach — a different

model, agent pattern, or architecture for a different need — rather than

forcing everything through a single template.

• Live on the bleeding edge. Actively track new models, agentic frameworks,

techniques, and tooling as they emerge; prototype quickly, form a view,

and bring the ones that matter into production with sound judgement about

what's ready and what's hype.

• Deliver, dependably. Ship software to production on time and to a high

standard, navigating the controls, governance, and compliance

expectations of a regulated environment.

• Embed with the business. Work shoulder-to-shoulder with stakeholders,

shape the roadmap, and translate business goals into working AI

capability.

• Set the standard. Define how agentic AI is built here — quality,

evaluation, observability, and responsible-AI practices that others

follow — and be the trusted technical authority for AI delivery across

the organisation.

ESSENTIAL (MUST-HAVE)

• Building AI/LLM applications in production — real systems people rely on,

whether internal or customer-facing, not prototypes or demos.

• Taking ideas from prototype to production — a proven ability to move from

a fast proof of concept to a hardened, scalable, maintainable production

system, and the judgement to know what changes between the two.

• A proven track record of delivering software to production in a regulated

environment — real, shipped systems with lasting impact, delivered within

governance, risk, compliance, and audit controls (e.g. financial services

or similar) without losing delivery pace.

• Hands-on experience with agentic AI — building systems where models

reason, plan, call tools, and take actions, including multi-agent and

agent-to-agent (A2A) coordination, tool and data integration via the

Model Context Protocol (MCP), and the control and safety mechanisms that

keep autonomous behaviour trustworthy.

• Identity and access for AI — how agents authenticate, act on behalf of

users with delegated authority, and operate under non-human/service

identities, with the authentication, authorisation, and auditability a

regulated environment demands.

• Strong software engineering fundamentals — reliable, observable, testable,

maintainable systems.

• Excellent communication and stakeholder skills — fluent moving between a

business conversation and a technical design, and comfortable working

directly with non-technical stakeholders.

• A forward-deployed mindset — comfortable with ambiguity, close to the

business and the user, biased toward shipping and iterating.

STRONGLY DESIRABLE

• Retrieval-augmented generation (RAG) — grounding responses in trusted

organisational knowledge.

• Evaluation (evals) — defining, measuring, and improving the quality and

reliability of AI and agent behaviour.

• Guardrails and safety — input/output filtering, hallucination mitigation,

control over autonomous actions, and responsible-AI practices.

• Prompt and context design, memory, and other techniques for shaping

reliable model and agent behaviour.

• Choosing the right approach for the problem — range and judgement to

select between different models, agent patterns, and architectures as the

problem demands.

• Leading architecture across teams — setting technical direction that other

engineers build on.

• An insatiable pull toward new and emerging technology — you keep up with

the frontier because you want to, and you can tell production-ready from

hype.

• Experience building customer-facing AI — delivering AI systems that real

external customers interact with, and handling the extra bar for safety,

trust, and polish that comes with it.

• Exposure to agent-driven and AI-generated UI — interfaces that agents

render, drive, or adapt dynamically.

All Senior Leaders are expected to demonstrate a clear set of leadership behaviours to create an environment for colleagues to thrive and deliver to a consistently excellent standard. The four LEAD behaviours are: L – Listen and be authentic, E – Energise and inspire, A – Align across the enterprise, D – Develop others.

All colleagues will be expected to demonstrate the Barclays Values of Respect, Integrity, Service, Excellence and Stewardship – our moral compass, helping us do what we believe is right. They will also be expected to demonstrate the Barclays Mindset – to Empower, Challenge and Drive – the operating manual for how we behave.

Minimum Salary: $235,000

Maximum Salary: $285,000

The minimum and maximum salary/rate information above includes only base salary or base hourly rate. It does not include any other type of compensation or benefits that may be available.

_