zach.dev
StartupCities
StoresData
Hire Me
LearnClaude
Spotify CLI
BuiltWith MCP
Claudem Ipsum
Who?I'm a software engineer and product leader focused on modern web technologies and AI.
An angel investor once described my core soft skill as: "The ability to perceive exactly what needs to be done. And then to do it."
My experience working in difficult environments around the world means that I can be trusted to get things done, even when things go wrong.
This website was built by hand, before AI.
Here's an ongoing autobiography, which shares my story of leaving school in 9th grade to join the workforce: now the subject of a chapter on grit and resilience in the bestselling book Mindshift by Barbara Oakley.
I've won several awards for economic research and have been published or interviewed in Virgin Entrepreneur, a16z's Future.com, The Atlantic's CityLab, Foreign Policy, and in academic volumes by Routledge and Palgrave MacMillan.
In a past life, I coined the term "Startup Cities" as co-founder of StartupCities.org in Guatemala City and a startup spinoff, both of which focused on why startups should build cities. This was originally a fringe idea, but has now attracted significant capital and founder interest in the 2020s. I sometimes write about Startup Cities at StartupCities.com
I read and collect obscure memoirs from the history of business and technology. You can find some of them in my public library. Send me your recommendations!
Wait... what is this site?
This is my personal portfolio, inspired by the question: "What would the opposite of the two-color template developer blog look like?"
Click the Start Menu to learn more. Have fun!
Contact:hello @ zach.dev
Jev Is the Most Important Model Announcement This Year
If you do meat and potatoes AI automation, "Jev" is the most important model announcement this year. I'm psyched.
The dirty secret of AI automation is that LLMs are expensive, slow, and often unreliable for business workflows, especially at scale. You spend tons of time bullying LLMs to do simple tasks like extract text, categorize items, or make true/false decisions.
Jev is an architecture built for these "boring" automation tasks. It doesn't write code or chat with you. It's more like an old-school classifier from 2018 (shout out to the 2018-2019 FastAI CNN/RNN crew!)
The problem with old-school models is that they're annoying to train. You need time and relevant data, and a lot of fussy iteration. But Jev is "zero-shot", meaning you just send in data as context, ask the question, and get structured output with confidence probabilities.
Jev is a great example of the old software maxim "worse is better." Because the model is simpler and focused on automation, it's fast and insanely cheap.
If Jev works well in production, this is likely a more important innovation for automation work than so-called AGI models like Astra or Fable.
I plan to write more about it after I use Jev a bit.
