Forward Deployed / AI Engineer — I build agentic AI that works inside locked-down enterprise data, in regulated finance and on SAP.
- 🏦 Contract Gen AI lead at Deutsche Bank (2023 – present): multi-agent LangGraph assistant for control testing — evidence gathering, cited conclusions, human sign-off, evals and model-risk review.
- 🛢️ Before that, seven years as an Azure / Databricks data engineer at Shell and MetLife.
- 🔌 Building a specialism in SAP data for AI: getting S/4HANA data into Azure and Databricks, and connecting it safely to agents.
- 📍 London
| Project | What it shows |
|---|---|
| sap-p2p-copilot | MCP server giving an AI assistant governed, read-only access to SAP S/4HANA procure-to-pay data. Deterministic three-way match in code, approval-only actions, audit log, tests and evals, Parquet landing for Databricks. |
| sap-field-service-bapi-catalogue | Integration catalogue of five SAP PM/CS BAPIs for field-service automation, plus a runnable demo: notification, equipment and parts BAPIs with commit/rollback, dedupe, an MCP server and 17 tests in CI. |
| sap-equipment-agent-api | FastAPI service over SAP S/4HANA API_EQUIPMENT + API_MAINTENANCEORDER: equipment status, location, last maintenance and technician as an AI agent tool (Claude tool use + MCP), with a one-page client integration spec. |
| q-dispatch | AI-assisted hospital insurance claim dispatch: bill pre-audit, claim email, inbox parsing and copay clearance. |
Deterministic logic for anything involving money or compliance · the model explains, code decides · humans approve anything that changes a system of record · every call audited · evals before every change.
Python LangGraph LangChain MCP RAG Azure OpenAI Azure AI Search Databricks Unity Catalog MLflow Azure Data Factory SQL SAP S/4HANA OData