Christopher Grimble

Researched and established AI development patterns to guide teams across Defra in building AI systems. Evaluated and prototyped AI patterns including RAG, LLM-as-a-judge, AI-powered OCR, and agent swarms for government suitability. Built prototypes using PydanticAI, integrating with frontier model APIs, AWS Bedrock, and Ollama-hosted models. Developed an operational prototype to help Defra content providers identify and update outdated guidance on GOV.UK, reducing a 2-year manual process to hours.

Worked on a greenfield premium finance system enabling Zego customers to pay car insurance in monthly instalments. Designed and built RESTful APIs using Python and FastAPI for premium finance workflows including instalment scheduling and calculation. Integrated with Experian for credit checking and affordability assessment during customer onboarding.

Built a proof-of-concept system using AI to create responses to industry and government consultations. Split the work into frontend, datastore, and agent workflow services. Used Azure OpenAI and RAG to determine whether a client should respond, generate key points, and produce a draft summary. Implemented in Python with FastAPI, SQLModel, and LangGraph, hosted on Azure, built using Terraform.