I build systems that
model reality.
From constraint solvers, digital twins to LLM-powered agents — I turn complex, messy real-world problems into production software for defence, energy, and finance.
Projects
SimAgent
Describe what you want to simulate in plain English. An LLM agent configures and runs the simulation, then visualises the results.
MCP Solver
An MCP server that gives Claude access to OR-Tools constraint programming — formulate and solve optimisation problems through natural language.
Fleet Schedule Explainer
RAG assistant that answers questions about complex fleet maintenance schedules — turning dense operational data into plain-language insights.
Resilience Digital Twin
Simulation framework modelling operational resilience for UK financial institutions. Pitched to five major banks, secured £300k funding.
Financial Reasoning Agent
LLM agent that answers multi-step financial questions by decomposing them into reasoning chains and executable computations.
Production AI on AWS
Full infrastructure-as-code deployment of an AI agent — ECS Fargate, Bedrock, Terraform. The infrastructure is the project.
About
I'm Samuel — a Senior Simulation & AI Engineer at Deloitte. I embed directly with client teams in defence, energy, and financial services, scope their hardest problems, and ship full-stack production systems.
My background is uncommon: I've spent years building agent-based models, discrete event simulations, and constraint solvers — systems that model complex real-world behaviour. Now I'm applying that same thinking to LLM-powered agents and AI applications.
I studied Physics at Durham and Artificial Intelligence at King's College London. I hold SC security clearance and have delivered systems that inform multi-billion pound decisions.