The shape of it.

What this work looks like: decades of code across multiple stacks, deployed per-customer, with the technical debt to match. The job is a parallel set of moves — analyze the interconnected data systems and produce a roadmap for what to retire, wrap, and rebuild; restructure how QA operates; drive infrastructure cost down; coach senior leadership through the change.

Fractional CTO for a 45-person engineering organization in an established software company.

The pattern that holds across these engagements: modernization is maybe 30% technology and 70% organizational. Standards that exceed team capacity fail. Pull-based adoption beats mandate. The people who adopt first are rarely the people who need it most.

CTO at an early-stage startup building a market intelligence and decision-support platform for participants in a federally administered healthcare program. Large-scale regulated patient and claims data on one side. Product velocity on the other. The interesting work lives at that boundary.

In practice: end-to-end ownership of architecture, data engineering (cloud data warehouse, Databricks pipelines), application engineering, design, and the AI tooling embedded throughout. Small senior team. I write code, review code, make hiring calls, sit in customer conversations.

The pattern: zero-to-one in a regulated domain is a discipline about what you don't build more than what you do.

Three threads here, related by pattern more than by client.

At a fractional engagement: leading a product workstream to expose application capabilities via MCP (Model Context Protocol) — so customers using Claude Desktop, Cursor, or other LLM hosts query the platform directly. OAuth 2.1, partner integration tier, B2B distribution via Claude Team and Enterprise.

At Rise Gardens, as CTO: built and shipped an AI customer-support chatbot using AWS Bedrock, Pinecone, and LangChain. Reduced support ticket volume by 75%.

Underneath both: a public catalog of the infrastructure — behavioral testing for RAG/LLM apps, vector database population pipelines, cost-effective production architectures, security-isolated AI-assisted development environments. All linked from the writing page.

The pattern: AI is a product capability, not a feature. Treating it as a feature is how you get a chatbot nobody uses.

At Aginity: co-founder and CTO of a consulting firm building bespoke data and analytic systems for Fortune 1000 companies. Grew past a hundred people, $25M/year in service revenue, $20M in venture capital. Clients across retail, banking, insurance, supply chain, and media. Forecasting engines, customer analytics, data warehouse modernization, market segmentation.

Out of that work we built the leading SQL development tool for high-performance analytic databases — Netezza, Greenplum, Redshift, Snowflake. Deep technical partnerships with IBM, AWS, Snowflake, and Microsoft. The tool spun off as its own company.

At Rise Gardens: a data warehouse and analytics platform integrating fifteen-plus sources — application, IoT device telemetry, marketing, sales, supply chain, customer support, ads. CAC/ROAS/LTV/churn analytics, manufacturing-failure forecasting, cohort and segmentation analysis. Driving the operating decisions of the business.

The pattern: data is most useful when the people who need it can ask it questions.

Teams range from forty to a hundred-plus across this work. At Aginity, the consulting org grew past eighty consultants, with a forty-person product development team layered on top. At the fractional engagement: leading modernization across a forty-five-person engineering organization, onshore and offshore. Hardware-firmware-software team integration for IoT product development, distributed across the US, China, Taiwan, and Eastern Europe.

The recurring lessons: small senior teams do disproportionate work; team-building is mostly a coaching problem dressed up as a hiring problem; standards that exceed team capacity fail; offshore teams have to be active participants, not ticket-takers. Executive coaching has shaped how I lead — I apply those tools to the leaders I work with, often as much value as the technical work itself.

As an aerospace engineer I co-invented a variable compressor diffuser design for small turbine engines. Two US patents — 5,222,356 and 5,235,801. The engines built on that design operate in commercial regional jets.

At Rise Gardens, as CTO: led the design and shipping of roughly twenty thousand connected devices — custom electronics around the ESP32 microcontroller, AWS IoT Core, OTA firmware with canary deployment, FCC and UL certifications, supply-chain manufacturing across China, Taiwan, and Mexico. Concept-to-shipping in eight months for one product line. Multiple industry awards including a CES Product Innovation Award. Migrated the entire fleet from a server-based MQTT broker to AWS IoT Core mid-flight with no customer impact.

Hardware and software are different disciplines that share more than people think. The patience is different; the discipline is the same.