Toryo
A TypeScript coding-agent orchestrator with evidence-bound review, configured checks, and reversible Git checkpoints.
- 8 tool adapters
- Quality ratcheting
- Trust-based delegation
I build agent systems, evaluation tools, and full-stack products with TypeScript, Python, React, and Next.js. I'm looking for my next senior engineering role and sharing the decisions, evidence, and lessons behind my work.
Over a decade shipping software · 226 live-coding episodes · production AI engineering
Selected systems
Agent orchestration, a persistent Minecraft world, and a summarization experiment. Each case study explains what I built, how it works, and what remains unproven.
A TypeScript coding-agent orchestrator with evidence-bound review, configured checks, and reversible Git checkpoints.
Five local-LLM agents sharing a Minecraft world, with persistent memory, role-specific tools, and an experimental evaluation track.

A Python evaluation harness measuring fact loss across repeated summaries, with deterministic grading, committed transcripts, and corrected grader bugs.
Public record
For several years, I learned in public and taught the mistakes as well as the wins. That chapter helped shape my career and still influences how I make complex work legible to teams.
Field notes
Short engineering stories from long-running agents, model trials, and the bugs that only appear after everyone goes to sleep.
2 min read · ai-agents
Advancement 42. Mason crafted an oak sign, set it down by the stash, wrote on it, and rubbed a glow ink sac into the letters.
2 min read · ai-agents
Advancement 41. Mason set a brewing stand down beside the stash, stirred sugar into three water bottles, and took out Mundane Potions.
3 min read · ai-agents
Advancement 40. Mason crossed the Nether with a crossbow and 24 arrows, shot a blaze down inside the fortress, and picked up its rod.
Open channel
I'm interested in ambitious engineering roles, focused consulting, speaking, and collaborations where the system has to do more than look good in a demo.