At a Glance
- Tasks: Design and build cutting-edge AI systems on AWS, working with multi-agent pipelines and orchestration frameworks.
- Company: Join a forward-thinking tech company that values innovation and collaboration.
- Benefits: Enjoy 35 days leave, private medical insurance, and 40 hours of paid learning.
- Other info: Flexible working options available, with opportunities for hybrid and remote work.
- Why this job: Make a real impact in AI engineering while growing your skills in a supportive environment.
- Qualifications: Experience with Amazon Bedrock, Python, and deploying complex agent pipelines.
The predicted salary is between 60000 - 80000 £ per year.
A hands-on senior AI engineer embedded in client delivery teams, responsible for designing, building, and operating production-grade agentic AI systems on AWS. This role sits at the intersection of AI engineering and cloud architecture, owning the end-to-end implementation of multi-agent pipelines, knowledge bases, and orchestration frameworks using Amazon Bedrock and its surrounding ecosystem. The right candidate has shipped real AI agents to production—not just prototypes—and brings the rigour of a software engineer to a space that often lacks it.
Key Responsibilities
- Design and implement production agentic AI systems on Amazon Bedrock, including multi-agent orchestration, memory management, and tool integration using Amazon Bedrock AgentCore and Strands Agents.
- Build, integrate, and maintain MCP (Model Context Protocol) servers that expose capabilities to AI agents across client platforms.
- Architect and implement RAG pipelines using Amazon Bedrock Knowledge Bases, managing vector stores, embeddings, and document ingestion from S3 and other sources.
- Apply the A2A (Agent-to-Agent) protocol to enable interoperability between agents across systems and workflows.
- Instrument AI systems with observability and tracing tooling—CloudWatch, spans, and traces—to support debugging, performance monitoring, and compliance requirements.
- Integrate LLMs into client applications through prompt engineering, context management, and function/tool calling patterns.
- Leverage serverless infrastructure—AWS Lambda, DynamoDB, S3—to build scalable, cost-efficient backends for AI workloads.
- Collaborate with client engineering and product teams to translate requirements into agent architectures, contributing to technical roadmaps and AI strategy.
Skills & Experience
- Hands-on implementation of Amazon Bedrock agents, knowledge bases, and model inference in production environments.
- Direct experience designing and deploying agent pipelines with real orchestration complexity.
- Built or integrated MCP servers in a production or near-production context.
- Familiarity with Strands Agents and their application to agentic workflows on AWS.
- Knowledge base design, chunking strategy, vector store configuration, and retrieval evaluation for RAG implementation.
- Strong applied proficiency in Python in an AWS and AI context.
- Working knowledge of Lambda, DynamoDB, and S3 as components of AI system backends.
- Experience instrumenting AI systems with logging, monitoring, and tracing tooling (CloudWatch preferred).
- Prompt engineering, tool/function calling, context window management, and output parsing for LLM integration.
- AWS certifications desirable, particularly the AI/ML Specialty.
What we’ll offer you
We trust people to do their best work. That means flexibility over rigid rules, impact over activity, and real investment in your growth both professionally and personally. You’ll be part of a supportive, friendly culture surrounded by smart, curious people who care deeply about what they do.
Benefits:
- 35 days leave (including bank holidays).
- Private medical insurance.
- Enhanced parental and adoption leave.
- 40 hours of paid learning and development.
We offer flexible working, including hybrid and remote options. Our office hubs are located in Edinburgh, Leeds, Manchester, London and Bulgaria, with occasional travel to client sites or CreateFuture offices when needed.
