Engineering executive · founder · open-source builder
I build software, teams, and platforms around hard problems — lately, mostly AI agents, memory, developer tools, and agent-native systems.
My career has taken me from co-founding a startup, to building emerging technology inside a Fortune 50, to leading large engineering organizations. Through all of it, I've kept writing code.
I don't build to prove I still can. I build because building is how I learn.
I believe we're entering a new phase of software architecture.
For decades, we designed primarily for:
UX — interfaces for humans DX — interfaces for developers AX — interfaces for agents
AX doesn't replace UX or DX. It adds a new first-class consumer of software.
Agents need more than APIs. They need:
- durable memory and context
- semantic understanding of systems
- tools with clear capabilities and boundaries
- permissions and safety constraints
- feedback loops that improve over time
- protocols for agents to discover and collaborate with other agents
I’m particularly interested in a future where an API is more than a collection of endpoints. It exposes enough meaning for agents to understand what a system can do, negotiate how to use it, and coordinate with other agents.
That's the rabbit hole I'm exploring now.
Persistent memory for the Pi coding agent.
Instead of hiding memory inside an opaque database, pi-memory keeps durable knowledge in plain Markdown and layers retrieval on top.
Some of the ideas I'm exploring:
memory → context → retrieval → compaction → recovery → continuous learning
I care less about adding another vector database and more about answering a harder question:
What does an agent actually need to remember to get better over time?
Persistent memory for Claude Code, OpenAI Codex, Cursor, and other coding agents.
The goal is to separate memory from any single model or agent runtime:
Claude Code ─┐
Codex ───────┤
Cursor ──────┼──► agentmemory ──► durable context
Agent ───────┤
future agent ┘
Your memory should belong to you, not to whichever agent happens to be running today.
I'm exploring how this can evolve beyond individual sessions into hierarchical memory for things like repositories, teams, Slack channels, applications, and organizations — with a feedback loop that continuously improves what the agent remembers.
Founder / CTO — Knock
I co-founded Knock and built the early product and engineering organization from zero. It eventually became a front-office platform serving millions of multifamily rental units and was later acquired by RealPage.
That experience taught me the difference between building software well and building the right software.
Lowe's — Innovation, platforms & AI
Over seven years I worked across several very different engineering problems:
- built and scaled a 3D commerce platform
- led Lowe's Innovation Labs engineering
- modernized critical supply-chain systems
- built an early retail Custom GPT prototype
- later led Core AI and production AI initiatives
The recurring lesson was that good architecture isn't about using the most sophisticated technology.
It's about choosing the simplest system that matches the shape of the problem.
CVS Health — engineering at healthcare scale
Today I lead engineering in healthcare, where reliability, organizational design, AI adoption, and the consequences of software decisions all operate at a very different scale.
It keeps reinforcing something I've come to believe strongly:
Engineering leadership is ultimately systems design — software systems, organizational systems, and increasingly, agent systems.
Build to understand. A working artifact teaches me more than a hundred slides.
Prefer simple systems. Complexity has to earn its existence.
Design away heroics. The best operational system is the one that doesn't need someone brilliant awake at 3am.
Hire for trajectory, not pedigree. Raw ability, curiosity, ownership, and growth mindset compound.
Know when to move fast — and when to do it right. The hard part isn't knowing either approach. It's knowing which situation you're in.
Stay curious. I'm still influenced by the original hacker ethos: understand systems deeply, take them apart, build things, share what you learn.
If you're working on AI agents, memory, agent infrastructure, developer tools, or agent-native software architecture, I'd love to compare notes.