Senior Manager, Automation and AI at Synopsys. I build multi-agent systems that do real work in regulated enterprise operations, where being wrong is expensive and every decision needs an audit trail.
- Multi-agent orchestration for accounts receivable and cash application, with deterministic arithmetic and human approval gates
- Entity resolution for ambiguous payer identities, the hard part of every reconciliation problem I have worked on
- Agent observability and run tracing, because a multi-agent system you cannot inspect is a multi-agent system you cannot ship
- MCP gateway patterns for enterprise tool access: auth, routing, rate limiting, audit logging
- Leading a global automation and AI team across Canada, the US, the UK, and India
| Project | What It Does | Stack | |
|---|---|---|---|
| Ledger Sense AI | Five-agent AR cash reconciliation for ambiguous payments. 34 catalogued edge cases, deterministic arithmetic, human approval before any ledger write. | Python, GPT-5.6 | Repo |
| Cash Application Foundry | Five-agent Azure AI Foundry swarm for enterprise AR cash application. Built for the Microsoft Build AI Hackathon. | Python, Azure AI Foundry | Repo |
| DisputeIQ | Vendor dispute defense system built on UiPath Maestro case management. UiPath AgentHack 2026, Track 1. | Python, UiPath Maestro | Repo |
| AgentOpsLab API | Backend for multi-agent orchestration: run tracking, tracing, and observability for enterprise agent workflows. | Python, FastAPI | Repo |
| Healthcare Claims | Claims intake and adjudication support: document extraction, eligibility checks, denial triage. | JavaScript, Azure DI | Live · Repo |
| MCP Gateway Design | Reference architecture for an enterprise MCP gateway covering auth, tool routing, rate limiting, and audit logging. | Jupyter, MCP | Repo |
Languages
AI and Agents
Cloud and Data
Automation and Platform
I am always up for a conversation about agent architecture in regulated environments, or what actually breaks when you move a multi-agent system from a demo to production.