Dom Michalec
Dom Michalec›About

Hey there! Welcome to my corner of the web.

My name is Dom. I build products and the systems that support them. I'm currently working with the amazing team over at Test Double.

I've spent my career moving back and forth between technical and non-technical roles, so I'm as comfortable in the code as I am shaping the strategy behind what gets built, and I've learned to explain why it matters to an array of people who usually want different things from the same conversation.

I care about building things that people actually use, and about making the systems that support them maintainable over time. I also care about helping people understand how to work with AI, so they can build their own agents and systems.

Lately most of that work has been in designing agents that do real work, wiring them together into larger systems, and teaching people what I've learned so they can build their own.

rubypythonjs 
dom = Builder.new(
  focus:    ["product", "engineering"],
  stack:    ["ruby", "python", "javascript"],
  building: ["ai_agents", "multi_agent_systems"],
  teaching: "ai_fluency_workshops"
)
dom.ship

Engineering

I like to stay as hands-on as I can. I've shipped production systems in Ruby, Python, and a range of JS frameworks, from server-rendered web apps to interactive front ends. What I care about most is whether people find the interface easy to use, and whether the code and the architecture still make sense a few years after they ship.

  • Ruby: web applications, APIs, and the tooling that keeps them maintainable.
  • Python: data pipelines, backend services, and the glue code that makes AI systems work.
  • JavaScript: front-end frameworks, Node services, and the integration layer between them.

AI agents and multi-agent systems

I design and build agents that take on real tasks. In practice that means deciding which tools an agent can call, structuring its context so it makes good decisions, and adding the guardrails that keep it reliable once it's in production.

When one agent isn't enough, I architect the multi-agent version: an orchestrator that breaks the work down, specialists that each handle their slice, and the handoffs and evaluation loops in between that make the whole thing trustworthy.

How I think about agents

Start with the job to be done. Give the agent the smallest set of tools that can do that job, and measure the outcome before you add anything else.

AI fluency workshops

I run workshops that teach people how to actually work with AI: how to prompt well, how to decide what is worth automating, how to use their discernment and judgement, and even how to build a first agent. They're hands-on and built around the people in the room, and I've run them for entire engineering teams, product orgs, and operational leadership teams.

Recent writing

Notes on product, engineering, and working with AI. See all posts.

Dom Michalec›Contact

Contact

If you want to talk about product, engineering, AI agents, or running a workshop for your team, send a note below.

Send a message

0 / 250

Workshop inquiries

For AI fluency workshops, mention your team, what you're hoping they walk away with, and rough timing in your message. I'll follow up with some questions to find a format that fits.

© Dom Michalec