Phil Leach

Agentic AI Engineer building autonomous document processing pipelines and intelligent automation for metals manufacturers. Phil designs and implements AI agent systems that analyse, classify, and extract data from industrial documents — mill test certificates, material data sheets, and compliance records — reducing manual processing and improving data quality across the supply chain.

His current focus is the Agentic Document Analysis Pipeline at GoSmarter: a system that routes unknown or low-confidence mill certificates to an AI agent for structured analysis, sample accumulation, and automated training data generation. The pipeline uses GPT-4.1 with structured output validation to identify suppliers, extract field mappings, and build the labelled datasets needed to train dedicated extraction models — closing the loop between production classification and continuous model improvement.

Phil works across the full stack of the AI services platform: Python Container App Jobs on Azure, Service Bus message routing, Azure Document Intelligence custom classifiers, and the CI/CD pipelines that test, build, and deploy each service. He brings a methodical, research-driven approach to engineering — investigating established patterns before implementing, writing comprehensive tests, and documenting decisions so the team can continue without a phone call.

Before joining GoSmarter, Phil built experience in AI systems engineering and infrastructure automation, with a focus on making complex AI capabilities reliable and observable in production environments.