What working here is like
The team is single-digit. Every engineer ships product. Research is evaluated by what reaches users, not by what reaches preprint. The same person might write a CUDA kernel in the morning and review a session transcript in the afternoon.
We are remote-first across Europe and North America. We work in writing, asynchronously, with a small number of overlapping hours. We do not have managers in the traditional sense; we have peers who happen to be senior.
We are profitable, independently operated, and intend to remain both. That gives us latitude on what to work on and how long to take with it. We trade scale of headcount for depth of work.
No open roles right now
We do not have any roles posted at the moment. Nastia hires only when the work demands it. We would rather stay below the headcount we need than above the one we can mentor well.
That said, we always read inbound notes from strong engineers and researchers. If you have spent years on a problem that overlaps with our research (long-form memory, real-time inference, multimodal generation, social AI alignment), send us a paragraph describing what you have built and what you would want to work on next.
Email [email protected]. No formal application. A link to your work is enough. We answer when there is something to say.