#introductions
Say hello and tell us what you are building
AL
JR
YT
PA
Avatar
Five sections built around faces — a company directory, a testimonial wall, a proof band, a community hub and a threaded discussion.
Twelve people, three time zones
We hire slowly and keep teams small enough that everyone still knows what everyone else is working on.
Leadership3 people
AL
Amara Lindqvist
Chief Executive
DO
Daniel Okafor
Chief Technology
RB
Rina Baptiste
Chief Operating
Design4 people
YT
Yuki Tanaka
Head of Design
MK
Marta Kovač
Brand Director
TS
Theo Sørensen
Product Designer
NA
Nadia Aziz
Design Engineer
Engineering5 people
JR
Jonah Reyes
Principal Engineer
CE
Clara Esposito
Platform Lead
HM
Hugo Meier
Staff Engineer
PA
Priya Anand
Engineering Manager
LO
Lars Ødegård
Infrastructure
The reviews we did not write ourselves
We replaced four internal tools with this in a fortnight. The part I did not expect was how quickly the rest of the company stopped asking us for reports.
SV
Sara VollmerVP Operations, Cadence
The migration guide was accurate, which is rarer than it should be. We were live on a Thursday afternoon.
MB
Malik BrennanStaff Engineer, Fathom
Support answered in eleven minutes on a Sunday and the person who replied had actually read our logs.
EJ
Elena JørgensenCTO, Northshore
Our design and engineering teams finally argue about the right things now.
KA
Kwame AsareHead of Product, Lumen
AL
DO
YT
JR
CE
PA
MK
TS
NA
HM
LO
SM
+2kTwo thousand teams got here before you
From two-person studios to companies with a procurement department. The onboarding is the same either way.
MERIDIANNORTHWINDATLASLUMENFORGECADENCE
Come build in public
Eleven thousand people who use this daily, arguing about architecture and sharing what they shipped.
148 online
#show-and-tell
Ship something? Post it here
CE
MK
TS
#help
Stuck on something — someone usually knows
HM
LO
NA
SM
14 comments
Sorted by most helpfulThe third rewrite worked because we finally measured queue depth before touching the scheduler. Everything else followed from that one number.
This matches our experience almost exactly. We spent a month optimising throughput that was never the bottleneck.
Did you keep the old scheduler running in shadow mode during the cutover, or switch outright?
Bookmarking this for the next time someone on my team proposes a rewrite in the first week of a project.