I've been writing code for a long time. All up and down the stack. I was writing JavaScript before the major front-end frameworks — through Moo and prototype.js, through Backbone, through Ember, until we finally landed on React. I was writing React when configuring webpack alone required what felt like a computer science degree. I worked at AI startups before TensorFlow came out, and I remember how revelatory TensorFlow was, and then Keras. I also remember that 99% of the pain of doing AI back then was just getting CUDA drivers to install. The environments were horrendous.
I don't miss any of it. Not for one second.
I've started so many projects. Made so many prototypes. (ADHD zero-to-one brain — I'm telling you.) Which means I've done the plumbing so, so, so many times. Adding authentication to a Rails app once meant going line by line — writing it yourself or copy-pasting from Stack Overflow, just to make what you actually wanted to happen, happen. Then better frameworks. Then Auth0. They were all still frustrating to me — not because they were bad, but because the idea was alive and the plumbing was bleeding it out. Every config file, every environment error, every "why won't this CUDA driver install" — each one a small withdrawal from the account I needed full.
Here's the thing: I just needed this thing to exist in the world. That's it. The whole job description, as far as I'm concerned. I didn't want to be configuring my platform. I didn't want to be configuring my compilation, my environment, my authentication. I wanted to be having good ideas and bringing them to people whilst I was still lit up about them.
The exhaustion of not being able to do that — for someone whose whole engine runs on spark — was its own particular kind of grief.
And something I'm coming to realize, as more and more of my code shifts from human-written to human-reviewed to human-stewarded to human-has-better-things-to-be-doing-with-her-time, is that I don't miss code all that much either.
When I look back at fifteen years of this, the things you might expect me to mourn are exactly the things that haven't gone anywhere.
I've been going through the archives lately. Reminding myself what AI design looked like back then. Before the noise. Remembering how invigorated I was when I discovered category theory and spent years diving into it — finding ways to abstract design sensibilities in front-ends and pipelines in a theoretical way. There was a time I felt my primary art practice was painting with React itself. A weird micro-celebrity moment in a specific corner of a specific ecosystem.
But when I think about it honestly: it was never about typing the lines. The lines had to be typed. My excitement always started at the whiteboard.
The satisfaction was inventing. In tech, in all of it — always inventing. Code was just the material I had to work with.
The experience of using Replicate today versus standing up your own AI infra as a two-person prototyping operation: there's no comparison. How could I miss that world? I also don't miss deploying isomorphic server-rendered React apps before Vercel, before Zeit. These aren't losses. They're what it looks like when the plumbing finally gets out of the way.
As my process gets ten times faster, week over week. As my output climbs. As I look at what I've been bringing into the world lately and feel like abilities I'd already built to a senior level have somehow skyrocketed — I can't imagine going back. I've been doing this long enough to know what the real deal feels like.
I have skin in this game — just my own: what I'm doing with the rest of my career, how I'm showing up in this industry at this particular moment of it. I'm not selling you an expensive infrastructure platform or telling you to use my models. Sure, I'd love it if you put something in the tip jar, or subscribed to this newsletter I'm getting back off the ground. But mostly I'm just telling you what I see.
Compounding returns. Genuine compounding returns. For someone who has been doing this type of thing her whole life.
The spark is going where it's supposed to go.
She doesn't miss code.
And she is inventing more things than ever before.
I Don’t Miss Code and Inventing Things Isn’t Going Anywhere
I run on energy and sparkles. When an idea catches — when something wants to exist and I can feel it — that's the fuel. That's the whole engine. So the specific cruelty of the old way wasn't just that it was slow. It was that it was draining. And exhaustion is anathema to someone like me.