AI Product Engineer: Agentic LLMs for Enterprise Workflows

Magentic; Greater London; Hybrid; GBP 90,000 - 110,000; Full time

Benefits offered by this job

Competitive Equity

Enhanced parental leave

25 days holiday exc bank holidays + 1

In-office lunches provided

Salary sacrifice pension

Nursery schemes

Hybrid London HQ

Annual team retreat

Job summary

the company is seeking an AI Engineer to build the agentic core of its product in London. You’ll own end-to-end LLM-based features and agentic workflows, deploying models from the framework to ERP integrations for Fortune 500 procurement teams.

You’ll benchmark performance, safety and behaviour, and work with customers to tailor solutions. We value experience shipping LLM-based production systems and a deep understanding of language models, CS fundamentals, and Python-based feature delivery.

Qualifications

• Experience shipping LLM-based solutions to production.
• Strong understanding of language models and CS fundamentals.
• End-to-end feature development in Python.

Responsibilities

• Own development and deployment of LLM-based features and agentic workflows end-to-end.
• Develop new agentic frameworks for multi-agent collaboration.
• Benchmark LLMs for performance, safety and behavioral monitoring.
• Operationalise AI features by connecting them to customer systems.
• Collaborate with customers and internal teams to design optimal solutions.

Skills

LLM development

Python

Agentic workflows

Multi-agent collaboration

Stakeholder communication

Job description

the company is seeking an AI Engineer to build the agentic core of its product in London. You’ll own end-to-end LLM-based features and agentic workflows, deploying models from the framework to ERP integrations for Fortune 500 procurement teams.

You’ll benchmark performance, safety and behaviour, and work with customers to tailor solutions. We value experience shipping LLM-based production systems and a deep understanding of language models, CS fundamentals, and Python-based feature delivery.