Overview
The Agentic Engineer will engage in a six-month contractual role focused on developing AI-powered workflows for a prominent content and media client in the UK. This role emphasizes building reliable, observable, and efficient systems while documenting processes to ensure smooth handover to the client's permanent hires. The engineer will work closely with users and cross-disciplinary teams, leveraging AI to enhance coding, testing, and documentation efforts alongside their own contributions.
Responsibilities
- Map workflows with users and cross-disciplinary teams to define clear outcomes for agentic solutions.
- Design and build agentic systems, including orchestration, tool use, human approval, and failure recovery.
- Integrate AI capabilities with client platforms through APIs, events, and data services.
- Develop reusable components and patterns for the client's team to utilize.
- Implement automated evaluations and tests for task success, quality, accuracy, and safety.
- Establish observability, tracing, and enhancements for agent behavior, latency, and costs.
- Incorporate security, privacy, and content rights controls with built-in human review.
- Document decisions in a clear format for the client's permanent hires to adopt easily.
Requirements
- Proven expertise in software engineering, with a focus on clear, maintainable, and well-tested code.
- Hands-on experience building applications using LLMs and agentic patterns beyond basic prototypes.
- Experience with API integration, data services, and creating reliable end-to-end products.
- Ability to design automated evaluations and tests for AI systems, focusing on quality and safety.
- Familiarity with observability and production monitoring for AI-powered applications.
- Experience deploying and managing cloud-based services using modern CI/CD practices.
- Strong understanding of application security, privacy, responsible AI usage, and content handling.
- Demonstrated experience using agentic AI tools, directing, and validating AI-generated outputs.