Senior AI Engineer: Agentic Systems & LLM Orchestration

Centrica plc; United Kingdom; Hybrid; GBP 90,000 - 140,000; Full time

Job summary

Centrica plc’s Windsor-based Senior AI Software Engineer role focuses on designing and building agentic services on Bedrock and AgentCore Runtime, with multi-step orchestration, state handling, and robust error semantics to deliver reliable customer experiences.

You will own tooling, retrieval augmented generation, prompts as artefacts, latency optimization, safety guardrails, and production monitoring, partnering with governance and product teams to ship responsible AI at scale.

Qualifications

• Substantial software engineering experience with strong Python.
• Delivery of LLM or agentic systems into production for real users.
• Hands on experience with tool and function calling, multi-step agent orchestration.
• Production experience with retrieval augmented generation and answer grounding.
• Experience deploying and operating AI services in the cloud with CI/CD and IaC.

Responsibilities

• Design and build agentic services on Bedrock and AgentCore Runtime, including multi-step orchestration and memory management.
• Build and own the tool layer through which agents act, exposing business capabilities with robust error semantics.
• Develop retrieval augmented generation workflows, including content preparation, embedding, and grounding.
• Treat prompts as engineering artefacts: version, test, and track changes.
• Implement gatekeeping, guardrails, and safe escalation to human advisors.
• Instrument end-to-end behavior for observability with Datadog and QuickSight, to detect failures and improve reliability.

Skills

Python

LLM deployment

Agent orchestration

Retrieval augmented generation

Cloud deployment

CI/CD

Guardrails & safety

Debugging non-deterministic systems

Education

Degree in Computer Science or Engineering

Tools

Terraform

AWS Bedrock

AgentCore Runtime

Datadog

QuickSight

Job description

Centrica plc’s Windsor-based Senior AI Software Engineer role focuses on designing and building agentic services on Bedrock and AgentCore Runtime, with multi-step orchestration, state handling, and robust error semantics to deliver reliable customer experiences.

You will own tooling, retrieval augmented generation, prompts as artefacts, latency optimization, safety guardrails, and production monitoring, partnering with governance and product teams to ship responsible AI at scale.