I'm Justin Johnson. I Build Things.
Two decades in biotech. I went from the lab to running large data and AI teams. Strategy meetings, stakeholder management, roadmaps. Real work that helps get medicines to patients.
But that's not who I am.
I'm a builder. Always have been.
The Day Job
For twelve years I led AI at AstraZeneca, the last stretch running the AI Center of Excellence for R&D. I loved it. In August I join Tempus to lead applied AI: turning one of the largest clinical-data moats in the world into products clinicians actually use. I can't wait to start.
In between, I'm on sabbatical. Which for me means one thing: building full-time, for the first time in years, with no inbox to answer to. Right now that's two builds: the Builder-Leader podcast, and Agentry, a framework for running an entire department as a system of agents.
Leading at scale comes with the same trade everywhere. Your days fill with meetings, not code. You enable other people to build, and that's the right call. But the itch never goes away, and for a few months, I finally get to scratch it.
The Night Shift
So I build at night. Weekends. Early mornings before anyone's awake. On planes. On trains. In hotel rooms.
Not because I have to. Because I can't not.
What I didn't plan on: the night work kept leaking into the day. Tools I prototyped at 30,000 feet ended up in the day job. Patterns I worked out on a weekend became how a 120-person team operated. The two halves fed each other.
One I'm proud of: I built ARIA, a research agent that runs on my own hardware, generating and testing hypotheses around the clock. We pointed it at real work for SocialEyes, a team in Nepal building AI that screens for disease from a single eye scan, in clinics with no specialist down the hall. ARIA helped their small team move at a pace their headcount shouldn't have allowed. Something I built for myself ended up helping people I'd never otherwise have reached.
And now it scales. We just won an NVIDIA Inception grant to carry the SocialEyes work forward. ARIA earned my first patent, with an arXiv preprint on the way. This is what it looks like when the night build goes to scale.
Over 18 months that compulsion has produced more than 50 projects. The ones I keep coming back to:
ARIA: a research agent that runs 24/7 on my own GPU, generating and testing hypotheses while I sleep.
Personal agents: OpenClaw and a few homebuilt ones run a shared backlog without me in the loop. Hermes is Nous Research's agent, retooled into my investing partner.
The AI Wire: reads the AI firehose every day, scores it, ships the signal.
Claudelicious: the cookbook for the whole harness, so anyone can run the same playbook.
Headroom: a menu-bar gauge for how much of your AI coding subscriptions you have left.
Kindling: self-hosted personal finance, private by default, on your own LLM.
EmberPlan: a FIRE planner. Small sparks, big retirement outcomes.
MarkItDown for VS Code: converts almost any file to Markdown. Nearly 40,000 installs.
Each one taught me something. Each failure sharpened the next. The AI-collaboration skill compounds: what took weeks takes days, what needed a team needs me and a well-built system.
The Philosophy
We're living through the biggest change in how knowledge work gets done since the spreadsheet. Maybe bigger.
I call it the 1:N effect. One person, properly equipped with AI collaborators, produces what used to take a team. Not because AI replaces judgment. It doesn't. Because it removes the friction between having an idea and making it real.
Think about everything that used to sit between "I have an idea" and "here's the working thing": learning the framework, writing boilerplate, debugging syntax, hunting Stack Overflow, juggling a dozen docs tabs. AI collapses all of it. The gap between imagination and implementation shrinks to almost nothing.
This isn't about working harder. It's about leverage. The same effort, ten times the output, once you stop fighting the tools.
The Evidence
I wanted to capture this somewhere. Not a resume, resumes are dead. Not a LinkedIn profile, those are for recruiters. Something that shows the work and the connections between the projects.
So I built justinhjohnson.com. An interactive archive of the building. Every project is a node in a graph. Click one, see what it does, why it exists, what it connects to. There's a story mode, a timeline, a stack view. The site itself was built in two days with the same workflow it documents. The recursion is the point.
The Book
Here's what I didn't expect: doing all of this changed how I lead.
The skills you build collaborating with AI at night, intent, orchestration, judgment, knowing when to step in and when to stay out, are the same skills that make you better at leading people during the day. The builder and the leader stopped being two different people.
So I wrote it down. Builder-Leader: The AI Exoskeleton is the book about that arc, the one from "I can build anything" to "I can help anyone build anything." It's the manual I wish I'd had on the way up.
Preorder it at builder-leader.com.
What I'm Exploring Now
The frontier keeps moving. Where my attention is:
Agents that act, not just answer. The shift from AI-as-search to AI-as-colleague is the most important trend in the field.
Autonomous research. ARIA runs continuously, compounding knowledge over time. A living research organism.
Small, focused models. The bet: a small model trained on the right data beats a giant one trained on everything.
The content flywheel. Every experiment becomes a post. Every post becomes social. The pipeline runs itself.
The Invitation
If you're building with AI, I want to hear about it. The community pushing these boundaries is still small enough that we can learn from each other.
Find me:
Blog: rundatarun.io, technical deep dives and building-in-public
Book: builder-leader.com, preorder
Portfolio: justinhjohnson.com
X: @BioInfo · GitHub: @BioInfo · LinkedIn: Justin H. Johnson
The future belongs to builders. Let's build it.
By day I help bring better medicine to patients. By night I prove what one person can do with AI. The two are closer than you'd think, and the book is about exactly that.
Great to see this Justin - and 2026 is going to be a fantastic year with AI getting superb traction on solving problems. What a time we are privileged to live through! All the best - Dale