I build reliability primitives for AI coding agents.
My current focus is OpenCode: readiness checks, runtime maintenance, session review, local project memory, and multi-machine agent coordination. I spent 21 years building backend systems, data platforms, and distributed infrastructure. Now I apply that production engineering background to AI-assisted software development.
Small tools that make AI coding agents more reliable in real engineering workflows.
| Tool | What it does |
|---|---|
| opencode-waitfor | Adds wait_for for HTTP, TCP, and command readiness checks so agents stop writing fragile sleep loops. |
| opencode-db-clean | Safely analyzes and vacuums bloated OpenCode SQLite databases with backups and integrity checks. |
| opencode-session-reflection | Reviews agent sessions as engineering evidence: communication gaps, recurring failures, and reusable workflow opportunities. |
| opencode-plugin-memory | Gives OpenCode a local, inspectable project-memory layer with selective cross-session recall. |
| opencode-fleet | Lets one master OpenCode instance coordinate multiple remote OpenCode nodes across machines. |
21 years in software engineering across backend infrastructure, big data, distributed systems, and AI-native applications. Earlier work included Hadoop, Spark, Flink, Hive, Kafka, and production data platforms. Recent work focuses on LLM applications, multi-agent systems, LangGraph, OpenCode, and AI-assisted engineering workflows.
OSS contributor to Byzer, Apache SeaTunnel, and Apache DolphinScheduler.