I'm a Software Engineer focused on AI systems, building the runtimes, control planes, SDKs, and orchestration layers that make model-driven workflows usable in practice.

Most of my recent work is in Go and TypeScript across agent harnesses, multi-agent orchestration, protocol integrations, and developer-facing tooling. I tend to build with a strong bias toward typed configuration, testable interfaces, runtime observability, and operationally clear systems.

• AI runtime and harness development
• Agent orchestration and workflow automation
• MCP and A2A protocol integrations
• Web and CLI control planes for long-running AI workflows
• Developer tooling for safe, observable AI systems

• sigil - a Go-based RLM harness with example-driven configuration, indexed run summaries, and token accounting surfaces.
• sigil-web - a TypeScript web UI for the sigil app-server with live run inspection, step-level views, and operator workflows.
• project-sigil - the superproject coordinating the sigil ecosystem, specifications, and multi-repo delivery flow.
• go-contextforge - a Go SDK for the IBM ContextForge MCP Gateway, covering tools, resources, prompts, gateways, and agent management.
• go-symphony - a Go orchestration service for running coding-agent sessions against issue queues in isolated workspaces.
• agent-skills - agent skills and plugin work that supports AI-assisted developer workflows.

My background in distributed systems engineeringshapes how I build AI-focused software today. I approach runtimes, agent orchestration, protocol integrations, and control planes as distributed systems problems: explicit interfaces, observable state transitions, fault tolerance, coordination boundaries, and infrastructure that remains understandable under real operational pressure.

• More about me