About
A platform for engineers, by engineers.
Codritium is small. We have one idea: what makes a senior engineer today is judgment when AI is in the loop. That skill can be taught, measured, and recognised.
2025
Year founded
September 2025, in Sydney, New South Wales
Remote-first
How we work
Distributed team. Async by default.
Independent
Funding
No marketing budget. No ads.
AU
Where we are
Sydney, New South Wales · ABN 68 872 276 630
Mission
Train developers to solve real engineering problems in the AI era.
The engineering job changed when AI joined the team. What counts as senior changed with it. We make that change explicit — and give working engineers the practice, the credential, and the words to operate inside it.
Senior engineers don't win by typing fast. They win by deciding what to accept, what to reject, and what to defend. That decision loop has always existed. Now it's visible enough to be taught. Codritium is built around it.
The thesis, in one sentence
The senior engineer of 2026 isn't the fastest typist. They're the fastest at rejecting bad AI suggestions — and the best at defending what they keep.
Principles
Four things we won't compromise on.
We measure judgment, not lines
Every challenge is scored on three things: defensibility, regression rate, and time to a correct fix. Not commit count. Not prompt count. Not whether you used AI.
Replay is a first-class artefact
The conversation. The redirects. The rejections. That's the work senior engineers have been doing in silence. We make it reviewable.
Research, not marketing
We publish methods, raw rubrics, and benchmarks anyone can replicate. We don't publish hype. If others can't reproduce a finding, it doesn't ship.
Working engineers run the panels
Every reviewer is a working engineer with the credential. We pay for review time. We rotate panels. We publish how often reviewers agree, every quarter.
Founder
Built by Ray Li, in Sydney.
Ray Li founded Codritium in September 2025. The idea came from a simple pattern: AI changed how code gets written, but how we judge engineers hadn't caught up.
Founder note · Ray Li
I kept seeing the same scene. A senior engineer would quietly reject an AI suggestion, redirect the model, and ship the right fix. None of that judgment showed up in code review, performance cycles, or hiring loops. It was invisible. So it didn't count.
The platform makes that loop visible. The research makes it measurable. The certifications make it portable. Everything else follows from that.
What Codritium is not
Three things we don't want to be confused with.
People mix us up with these three. We're none of them.
Not a bootcamp
Bootcamps teach you how to start. We teach you what makes you senior — judgment when AI is in the loop.
Not a LeetCode clone
Algorithm puzzles aren't what holds engineers back at work. Real bugs are. Security drift is. Defending a refactor is.
Not a generic AI-education platform
We don't teach prompting. We teach you to make engineering decisions you can defend when AI is in the loop.
Roadmap
Where we are. What's next.
Codritium is early. We publish what's shipped and what's queued, in order, so you can plan around it.
Now
Public site, brand, research thesis
What you're reading. The base for the research program and the challenge platform.
Next
Challenge platform — closed beta
Challenge runner. Replay capture. Official solutions. First proctored certification cohort.
Then
State of AI Engineering · 2026
Our first benchmark of human + AI engineering pairs. 12 task categories. Public methodology. Open to replication.
Later
Books, contributor program, competitions
Engineering books. A program for working engineers to author challenges. Weekly timed rounds with panel review.
Get involved
Three ways to join in.
We grow through contribution. No marketing budget. Our community is how people find us.