I build with AI.

And write about what actually works.

Software engineer at the National Cancer Institute. Master's in AI/ML. Spent 8 years building and running a factory before writing my first line of code. Now building multi-agent systems. Celebrating the wins, learning from the failures, and writing the manual as I go.

Read the latest → See what I'm building
Writing All posts →
  1. May 2, 2026

    Memory, Routing, and Bounded Ownership in Agent Teams

    Why agent teams scale through scoped memory, specialist lanes, explicit routing, and operator-visible handoffs, not by giving every agent more context.

    Agentic Development · OpenClaw · Multi-Agent Systems · Engineering
  2. April 30, 2026

    Operator Control for Agentic Systems

    Why I am building OpenClaw around visible work, clear ownership, approvals, and intervention.

    9 min read · Agentic Development · OpenClaw · Product Design · Engineering
  3. April 13, 2026

    I Built an AI Email Pipeline That Sends 450 Emails a Week. It's Manual on Purpose.

    Everyone assumes AI plus email equals full automation. I went the other direction. Here's why manual-first beats hands-off for B2B outreach, and how the pipeline actually works.

    12 min read · ai-ml · engineering · build-in-public · python
Building All projects →

OpenClaw

Active

A founder-OS control plane I'm building around operator control: visible task state, approval guardrails, durable multi-agent workflows, and a single inbox for the work that needs my attention.

AI/ML · Agentic · Operator Control · Node.js

Agentic Development System

Active

A spec-driven, multi-agent system for test-driven development across production JavaScript codebases. Works with Claude Code, Gemini CLI, and Codex.

AI/ML · Agentic · JavaScript · TDD

CourtMix

Rebuilding

A local racquet-sports matchmaking app, rebuilt ground-up from Racquet Rivalry (2022) on my agentic development system.

Flutter · Firebase · GCP · Mobile

RL Trading Environment

Research

Research on applying deep reinforcement learning to EMini futures trading. Cross-attention models, custom reward engineering, and a lot of humbling results.

Python · PyTorch · Reinforcement Learning · Finance
Now Updated May 2026 →