sharing beauty
Engineering and developer-experience leader
I build developer tools, knowledge systems, and workflows that help people and AI agents use complex technology. At Aztec, I connected versioned documentation, executable tools, and human feedback. I am exploring how that approach changes what we can learn and do when AI takes part in everyday work.
This site is organized around active inquiry. Some questions may become projects or notes; others are useful because they remain open.
Professional work, personal tools, and unfinished investigations belong here together. Each is a way of testing an idea in the world.
Featured work
I designed AI-assisted support workflows connecting models with versioned documentation, source repositories, community knowledge, and executable development environments. The system used uncertainty and developer failures as signals for human validation, documentation fixes, and product feedback.
I built and tested tools for helping coding agents work with unfamiliar systems: an agent benchmark, an MCP server, a Claude Code plugin, and a reproducible development environment. Together they explore how context, tools, and evaluation change developer experience when agents become capable builders.
Agent Builder Bench (external site)MCP server (external site)Claude Code plugin (external site)Development environment (external site)
A reading application with an S3-compatible Cloudflare R2 storage layer for uploaded documents. I worked with object uploads, streaming, deletion, and a script for copying files between production and staging buckets.
A personal reading project exploring what a book can become when notes, questions, and patterns stay connected. An opt-in intelligence layer moves selected reading events through Cloudflare Workers and Databricks to build reader-specific concepts and recommendations. That work is still under review.
An experiment in fine-tuning small models for wallet tasks, with local-device use as the goal. I built training and evaluation workflows to test whether a model can propose useful actions while deterministic code retains control of approval and signing.
As Head of Developer Relations, I led a six-person team helping developers adopt a privacy-focused smart-contract platform and the Noir programming language. I worked across documentation, examples, workshops, support systems, product feedback, and technical strategy.
Developer Relations strategy (external site)Aztec (external site)Noir (external site)Build with ZK workshop (external site)Wonderland CTF walkthrough (external site)
Other work and experiments
An exploration of private, on-chain transfers using zero-knowledge proofs and homomorphic encryption. It asks what programmable privacy can feel like in an application, rather than treating privacy as a purely infrastructural feature.
A native Android proof of concept for generating and verifying Aztec client-side proofs on the device. It explores what private application infrastructure might look like when proving, verification, and eventually wallet flows can move closer to the person using them.
A hackathon privacy application suite for ERC-20 tokens, including sealed-bid auctions, private crowdfunding, and private token voting. It was an experiment in giving people more agency in the everyday applications built on public blockchains.
Personal and emerging experiments
An experiment in using AI as an interpretive partner, not to deliver authoritative answers, but to help people explore recurring symbols, patterns, and perspectives in their inner lives.
An earlier experiment in generative identity, onchain art, and composable visual systems.
A planned workshop on agentic fundamentals beyond packaged products: terminal-based harnesses, always-on home servers, Raspberry Pi systems, Telegram interfaces, home automation, and multiple persistent agent sessions.
Using PianoBooster while learning piano and music theory, with an interest in the feedback loops that make sustained practice more legible and rewarding.
Essays, provisional notes, and open questions. The shorter pieces may remain seeds; the longer ones follow an idea through the work that shaped it.
Agent design optimizes a part. What a system can accomplish lives in the arrangement around it: humans, machines, knowledge, tools, and feedback.
LLMs accelerate both understanding and action. As the cost of producing software falls, discernment and evaluation become the bottleneck.
What I learned building versioned knowledge, agent-readable documentation, and execution environments for developer education and support.
How agent workflows turned developer questions and failures into changes to documentation, tooling, and product development.
Documentation is increasingly used by more than a human reader. I am interested in what makes technical knowledge legible to people and capable agents without reducing it to a pile of retrieval fragments.
Good documentation carries context: why an interface exists, how it fails, what has changed, and where uncertainty remains. Agent-readable systems may need the same qualities, expressed in structures that can be queried, tested, and kept current.
Privacy is not only about keeping secrets. It is part of the practical space people need to experiment, change their minds, form relationships, and make choices that are not immediately optimized by a surrounding system.
How might we design systems that help people notice, hold, and learn from complexity rather than prematurely simplifying it into control?
I'm drawn to books that explore agency, consciousness, technological change, human development, privacy, and the emergence of new forms of order.
I am exploring remote roles in AI engineering, developer experience, developer relations leadership, and forward-deployed engineering.
Applying AI inside organizations by improving existing workflows, identifying new ones the tools make possible, and helping developers understand complex software systems well enough to build with them.
Continuing to test small, private models after a bounded fine-tuning experiment for Agentic Wallet. I am learning where training helps, where it fails, and what local hardware can realistically support.
How people preserve judgement as agents accelerate learning, implementation, and the rate at which decisions spread through an organization.
Continuing a Detroit-area workshop series on agentic fundamentals and practical systems.
I am based in the Detroit area and have spent eight years helping developers engage with emerging technology, most recently as Head of Developer Relations at Aztec.
Head of Developer Relations, leading developer education, tooling, documentation, support systems, and AI-assisted developer workflows for Aztec and Noir.
Developer Relations Engineer designing developer experiences, technical education, example applications, and community programs for a mobile-first financial platform.
Ethereum Educator designing and operating technical curriculum for software developers entering the Ethereum ecosystem.
My recent work has focused on deploying AI across organizations, improving workflows, and helping developers learn complex systems. I am especially interested in what happens when production becomes faster than our ability to evaluate it. Outside of work, I keep returning to philosophy, complexity, collaboration, dreams, music, education, and technologies that support agency.
Across these different domains, I keep returning to the same question: what conditions allow greater intelligence, agency, and complexity to emerge?
[email protected]
GitHub (external site)
·
LinkedIn (external site)
·
Résumé
·
X (external site)
My interest in privacy began with the cypherpunk conviction that people need systems that preserve autonomy and meaningful choice. That concern feels more, not less, important in a world increasingly mediated by intelligent machines.
"We the Cypherpunks are dedicated to building anonymous systems. We are defending our privacy with cryptography, with anonymous mail forwarding systems, with digital signatures, and with electronic money."
Eric Hughes, A Cypherpunk's Manifesto (external site)