How we build clinical AI that clinicians can act on. Honest writing from the people doing the work: the systems, the trade-offs we made, and the problems we are still figuring out.
Every post is written by the engineer who did the work. No comms polish, no vague generalisations. These are the kinds of posts we publish.
The problem, the decision, the outcome, and what we would do differently. Architecture with the reasoning left in.
One hard problem, followed all the way down, with real system names, real numbers, and the implementation detail that matters.
What broke, what we did about it, and what changed afterwards. The fullest expression of the honesty we hold ourselves to.
The practices and values behind the code. How we handle quality, how we run incidents, how we hire. Only we can write these.
A real project shipped, told by the person who shipped it. What it is like to do meaningful work here from day one.
Lyrebird uses Lyrebird every day. What running our own product in the clinic taught us, and how it changed what we built.
We write about the parts most teams keep private: the trade-offs, the things we got wrong, and the decisions that only make sense once you can see the whole system. Clinical AI has to be trustworthy enough to act on. That standard starts in the code.
A single honest post can tell a strong engineer more about a team than any job ad. That is the bar we hold ourselves to here: would someone we would love to work with find this well worth their time?
We are hiring engineers who care about getting the hard parts right. If the way we build resonates with you, we would like to meet.
