Technology
Tools thatfit the problem.
Open by default, license-safe, and backed by deep team experience - never resume-driven development.
Tech plates orbiting an orange coreCategories
Our stack, by layer.
Stack
Representative stack by category.
| Category | Primary | Also in use | License stance |
|---|---|---|---|
| Frontend | React + Next.js | Motion, Embla | MIT - open |
| Backend | Node.js | FastAPI, Kafka | MIT/Apache - open |
| Mobile | Flutter | React Native, Swift/Kotlin | open + native SDKs |
| Cloud & DevOps | OpenTofu | Kubernetes, GitHub Actions | MPL/Apache - open IaC |
| AI & Data | Python | pgvector, LangChain | open; models benched |
| Databases | PostgreSQL | Redis, ClickHouse | open source first |
open by default; any non-OSS tool is a deliberate, documented exception
How we choose
Stack selection criteria.
open by default
we reach for open source; closed tools are the exception
license-safe
no BUSL traps; we pick open forks when licenses sour
community health
active maintenance, real adoption, clear governance
team depth
we ship what we know deeply, not what is trendy
performance
benchmarks decide; budgets enforced in CI
exit-friendly
no vendor lock-in; your stack stays portable
Technology FAQ
Questions, answered.
Why open source first?
Portability, auditability, and no license rug-pulls. Closed tools are deliberate exceptions.
Do you follow trends?
No - resume-driven development is banned. Team depth and benchmarks decide.
What if a license changes?
We migrate to open forks - e.g. OpenBao over BUSL Vault, OpenTofu over Terraform.
Can you work with our existing stack?
Yes - we integrate and improve; rewrites only when justified by evidence.
How do you pick AI models?
Task-based bench - open-weight and frontier; the scoreboard decides.
Get a stack that fits.
We pick the stack that fits, and we justify every choice.