AI engineering
- MCP
- Agent / SubAgent
- RAG
- Prompt Engineering
- Ollama
15 years in SAP and enterprise systems across development, implementation, integration, operations, customer success and project delivery. Now bringing AI engineering into those systems.
My career did not start with algorithms; it started with enterprise systems. In 2011 I was writing ABAP and running FI/CO operations, then SAP HCM/HR implementations, SuccessFactors and BTP CPI integration, and later AMS operations for multinational clients. I have seen how systems are designed, how they go live, and how they get fixed when something breaks at 2 a.m.
In recent years I have brought AI engineering back into that path: wrapping SAP into MCP servers so multi-role agents can analyse, change code, run tests and verify deployments inside controlled boundaries; connecting RAG, local models and prompt constraints to real workflows. AI is not another slide topic — it has to enter the system, the permission model and operations.
That is how I practise Forward Deployed Engineering: not advising from documents, but working inside the business and the system, producing the result, and handing it over.
Deliver and operate SAP HR, SuccessFactors and BTP integration for multinational clients; led the SAP × MCP AI agent platform that brings AI into delivery and operations workflows.
Managed Oracle Fusion Cloud enterprise accounts across China and Korea, coordinating global support and implementation around cloud strategy, KPIs and QBRs.
Core consultant for SAP HCM and E-HR Portal across Greater China and APAC — requirements, blueprints, configuration, custom development and training.
Delivered and operated a group-wide ERP, owning business blueprints, functional/technical specs, development, testing and go-live.
Ran FI/CO operations and advanced ABAP development for a group ERP, supporting overseas business lines and a Korea HQ handover.
Computer Software · B.Eng.
This page is intentionally a summary. The full CV, project details, metric definitions and references are available on request.
Wrapped SAP data dictionary, functions and transport requests into an MCP server, connected to AI coding assistants, and built a multi-role agent pipeline from analysis to code, tests and deployment checks.
Connected an AI orchestration engine to CPI through enterprise APIs, turning iFlow configuration and mapping into a loop of contract parsing, flow generation, one-click deployment and automated verification.
Designed the enterprise integration blueprint around Employee Central, connected dozens of branches through BTP CPI, and completed a compliant migration and interface rebuild for sensitive HR data.
Owned high-concurrency ERP operations and enhancements, leading CS after-sales, engineer dispatch, membership billing reports, interfaces and enhancements.
Additional experience includes a Thailand finance-process programme, SuccessFactors rollouts and Concur integration — available to discuss.
If it cannot be monitored, rolled back and handed over, it is not done.
Write actions are explicitly gated, sensitive data stays out, and AI gets no extra privileges.
Define the baseline first, then claim improvement — with numbers you can review.
Architecture, operations and incident handling all leave reusable documentation.
I own Chinese, Korean and English directly; Japanese is delivered with project partners. A four-language site is content coverage, not a claim of four native languages.
This page is intentionally a summary. The full CV, project details, metric definitions and references are available on request.
Tell me the current state, the outcome you need and your timeline. I reply within 24–48 hours and start with whether it is worth doing — and the fastest path.