00Rated capacity
What we lift.
How we work.
Four kinds of job, one way of running them. If your project fits below, we can probably ship it. If it doesn't, we'll tell you on the first call and save everyone a quarter.
01Service
Platforms
The core systems a business stands on: trading and clearing, booking and billing, policy and claims. We build them to carry load for a decade — boring schemas, honest APIs, migrations that don't require a war room.
Typical shape: replace or rebuild a load-bearing legacy system without stopping the business that runs on it. Shadow traffic, parallel runs, cutover windows measured in minutes.
- Typical duration20–40 wks
- Reference jobMeridian Clearing
02Service
Realtime
systems
Order flow, pricing, dispatch, telemetry — systems where latency is a feature and backpressure is a design decision, not a surprise. We work in microsecond budgets when the job calls for it and say so plainly when it doesn't.
You get published latency targets in the statement of work — p50, p99, p999 — and a load harness that proves them before cutover, not after.
- Typical duration14–30 wks
- Reference jobFerrostar Logistics
03Service
Data
engines
Pipelines and stores that make a terabyte an hour feel boring. Exactly-once where the money moves, at-least-once where it doesn't, and a bill your CFO can read. We name the trade-offs out loud instead of hiding them in a vendor diagram.
We are opinionated: Kafka for the log, Postgres until it genuinely can't, object storage for everything cold. Exotic databases need to win an argument first.
- Typical duration12–24 wks
- Common stackKafka / Postgres / S3
04Service
Rescue
missions
The project is late, the vendor is gone or should be, and someone senior is asking pointed questions. We stabilize first — incidents down, deploys safe, alerts honest — then rebuild along the system's real fault lines. In that order, always.
Expect blunt findings in week one, measurable calm by week four, and your own team running the system when we leave. We build the handover into the plan from the start.
- Typical duration8–16 wks
- Reference jobCastellan Insurance
05How engagements run
Same crew,
start to finish.
-
Fixed team
The 3–5 engineers who scope the job build the job. No rotation, no substitutions, no "resourcing action" halfway through. You'll know everyone's name by week two.
-
Code in your repo, day one
We work in your GitHub org from the first commit. Your CI, your review rules, your license. If we disappeared tomorrow, you would lose a vendor, not your codebase.
-
Weekly demos, working software
Every Friday you see the real system run — not a slide about it. Feedback lands in next week's build. You will never go two weeks without seeing working software.
-
Milestone pricing
Fixed price per milestone, defined in working software. You approve each one before the next starts, and you can stop at any milestone boundary and keep everything shipped so far.
06Refused loads
What we
don't do.
-
Staff augmentation
We don't rent engineers into your org chart by the hour. If you need staff augmentation, a hundred firms will oblige. We ship defined systems.
-
Maintenance-only contracts
We hand over systems your team can run and we build the handover into the plan. Keeping you dependent is a business model — just not ours.
-
Strategy-only work
No strategy phases, no 90-page target architectures, no roadmap exercises. If an engagement wouldn't end in running software, we decline it.
-
Impossible dates
If your date is impossible, we say so in the first meeting and show the math. We'd rather lose the deal than ship the excuse.