Custom software, telecom systems and computing infrastructure. Registered activity: NACE 62.10, computer programming.
Software for networksthat cannot affordto drop a packet.
We write low-latency backends, carrier signaling software and the clusters they run on. Senior engineers only, judged in milliseconds.
Services · 4 disciplines
What we build.
One team from the instruction set to the data centre. No handoffs between the people who write the code and the people who run it.
Production services in Rust, Go and TypeScript with fixed memory budgets and predictable tail latency. We write the hot path, the tests that prove it and the dashboards that watch it.
SIP and WebRTC relays, anycast control planes and traffic shaping that keep calls up when a regional link degrades. Built and load-tested against carrier traffic profiles.
Multi-region Kubernetes and bare-metal topologies with workload isolation, predictable storage IOPS and a cost per compute cycle you can read on one dashboard.
Senior engineers read your code and your traces, find the concurrency bottlenecks and single points of failure, then hand you a ranked list of fixes with effort estimates.
Track record
Measured, then written down.
Numbers from systems in production today. Each one comes from client dashboards, not from a benchmark on a laptop.
p99 ceiling held on production APIs we built
Measured uptime across carrier routing clusters
Moved between clouds with zero read downtime
Average compute spend removed after an audit
Every millisecond is part of the contract.
Latency budgets go into every statement of work.
Load tests run at three times your recorded peak.
A rollback path exists before the first cutover.
Selected work · 5 systems
In production now.
Open any system for the architecture, the before and after, and the stack.
Stack
What we ship and operate.
Chosen per workload. If a tool is on this list, someone on the team has been paged for it at 3 a.m.
Rust / Go / TypeScript / C / eBPF / DPDK / Linux / Kubernetes / Firecracker / Terraform / Envoy / WireGuard / PostgreSQL / ClickHouse / Kafka / Arrow / Raft / gRPC / SIP / WebRTC / BGP / Prometheus / Vitess / NATS / Nix
FAQ
Before kickoff.
The questions we hear on every first call. Anything else goes in the brief below.
Yes. Most projects run inside your repositories, CI and on-call rotation. We pair with your engineers so the system stays maintainable after we leave.
Fixed-scope packages range from €2,400 for a two-week systems audit up to €4,990 for core programming and telecom modules, with latency and uptime targets written into the contract.
Yes. We start with a 72-hour stabilisation pass: tracing, rate limits and rollback paths. Rebuild decisions come after the incident is contained.
Remotely from the EU, with overlap for North American hours. We travel for cutovers and data-centre work.
You do, from the first commit. Everything ships with runbooks, dashboards and architecture notes.
