A young team, taken on for the layer that has to be right.
Data-intensive software fails at the seams. We stayed small so there are fewer of them.
Small on purpose.
Systems that move a lot of data rarely fail in the middle of a component. They fail at the seams — between the ingestion and the computation, between the computation and the screen. Most of those are handover failures, not engineering failures.
So we stay small, and hire for depth rather than headcount. The people who scope an engagement are the people who build it, and the same team operates it until the handover is finished. There is no bench, and nobody is rebadged onto your project.
That limits how many engagements we can take at once, and we would rather say so than staff around it. It also means the person answering your question wrote the code.
One project at a time.
We take one engagement, and everyone works on it.
No parallel projects
One client has the team's whole attention. Nothing waits behind a second contract.
The people who scope it build it
There is no bench, and no handover from the people who won the work to the people who do it.
Depth is a scheduling decision
Quality comes from not dividing attention. We would rather start later than start thin.
Then we hand it over
Monitoring, documentation and a runbook, so the next team runs it without us.
The product we are building now is under Pre-A investment review by OKX Ventures.
The engagements we do best.
Work where the data is the risk: many sources, high volume, figures that have to reconcile, and a number somebody will eventually be held to.
Work where one team should own both the pipeline and the interface, because the two cannot be specified independently without something being lost between them.
Work that will be operated after we leave, which makes documentation and monitoring part of the build rather than an afterthought.
Where we work.
We work from the fifth floor of a new building at 12 Teheran-ro 107-gil, Gangnam-gu, Seoul — three minutes on foot from Samseong Station, Exit 8.
If the data is the risky part, say so.
An unfinished specification is fine. Business, investment and hiring all reach one address.