Evidence, not adjectives

The record.

The clearest evidence of design judgment is what survived production. Shipped AI systems, open-source infrastructure other engineers depend on, and a research-grade ML edge. Every number is real.

1,316: Commits on a production AI-agent SaaS, web + desktop.
663: Test files in that same shipped system.
58: Tools in reaper-mcp, our MCP server on PyPI.
47★: GitHub stars on reaper-mcp.

Shipped systems

Built, deployed, in users' hands.

Autonomous AI content department

One chat brief becomes finished, published multi-platform content.

The harness thesis at department scale: one orchestrator delegating to ~14 specialist sub-agents through a shared file bus, with idempotent recovery, eval-gated routing, human approval gates.

1orchestrator, ~14 sub-agents
Eval-gated model routing
Idempotent re-run recovery
Per-piece cost tracing
On-prem observability
Human approval gates

Orchestrator

Shared file bus

~14 sub-agents

eval-gated routing
human approval gates

Production AI-agent SaaS

A conversational, multi-tool AI agent across web and desktop.

A full-stack AI product on a single-coordinator, multi-tool agent, web and desktop. Stripe subscriptions, AWS, full observability — the full-stack AI SaaS pillar in production.

1,316commits
663test files
16backend modules
12+API integrations
5platforms

HIPAA-adjacent capture system

On-device transcription where the audio never leaves the machine.

A clinical-notes capture system, compliance from day one. Transcription runs on-device with local ML, so data never leaves the machine — macOS System Extensions and XPC, a sandboxed, crash-isolated build.

On-device ML, no data leaves the machine
macOS System Extensions + XPC
Audit logging, day one
Shipped

Document → structured extraction

Messy specs in, schema-validated data out — with citations.

Spec, quote, and proposal PDFs and DOCX in; schema-validated fields out, each citing the source page. Layout-aware parsing preserves tables; output enforced through the tool-use API.

Schema-validated output
Per-field source citations
Layout-aware parsing
Tool-use enforced

Open-source infrastructure

Tools other engineers run.

MCP servers and agent tooling we build in the open. Public, installable, used outside our work.

reaper-mcp

An MCP server giving AI agents live control of a target application. FastMCP, on PyPI.

58tools
47GitHub stars
On PyPI

vst-bench

An MCP server for AI-driven automated plugin testing. A Python server talks to a C++ test host over JSON-RPC with WebSocket streaming.

31tools
Python ↔ C++ over JSON-RPC
47unit tests

goalkeeper

An open-source Claude Code plugin for contract-driven, multi-agent goal execution. A subagent judge gates completion against an explicit Definition of Done, so a passing validator can never ship a placeholder.

v0.3.0
8skills
80lifecycle test assertions

clawd-cursor

A 40-tool desktop-automation MCP server (TypeScript, macOS). We contribute rather than author it — 26 commits, ~10% of the repo: macOS stability, GPU detection, accessibility.

40-tool MCP server
Contributor: 26 commits (~10%)

OpenClaw Live Events plugin

An agent tool and CLI integrating the Ticketmaster Discovery API, on NPM as `@itsuzef/live-events`. We are an OpenClaw org member with a merged upstream PR.

Published on NPM (@itsuzef/live-events)
Ticketmaster Discovery API
Agent tool + CLI

Where this work has gone

Industries served.

Real proposals and shipped engagements.

Real estate & proptech
Construction & AEC
Healthcare (HIPAA-adjacent)
Creator economy & marketing
E-commerce & retail
Fintech
Logistics & trucking
Hospitality
Podcasting
Robotics
ML research

Want the same rigor on your system?

We will tell you honestly whether we are the right team.

Start a conversation