THE WHOLE
SYSTEM, one TEAM
Most projects lose time where one supplier hands over to the next. We cover the whole path, from the equipment in the field to the screen someone makes decisions on, so there are fewer handovers to go wrong.
By the numbers
14m
DATA POINTS A DAY MOVE THROUGH WHAT WE BUILD
55 days median from kickoff to the first production release.
99.98% uptime across managed workloads, independently audited in 2025.
one team, whole path
NO INTEGRATOR IN THE MIDDLE.
NO SEAM TO ARGUE OVER.
[ HOW A PROJECT ACTUALLY RUNS ]
TAP A STAGE
stage 01
Something happens in the physical world and has to be captured reliably by equipment nobody visits often. Getting this right decides how much everything downstream can be trusted.
stage 02
Getting that data somewhere useful, over connections that drop. When it does, the system should know what it missed rather than quietly average over the gap.
stage 03
Turning raw data into a forecast, a warning or a number someone can act on, along with an honest statement of how confident it is.
stage 04
The screen an operator opens at 3am, the app a customer uses and the report a regulator reads six months later, all drawn from the same numbers.
[ WHAT WE TAKE ON ]
TAP A ROW · NO SUBCONTRACTORS
Batch and streaming in one lineage, so the report and the live view cannot disagree.
Schema changes arrive as contracts with owners, not as an incident on a Friday.
The evaluation harness is written first. A model nobody can measure is a liability with a dashboard.
Drift, coverage and confidence are reported to operators in their own units, not in F1 scores.
We sit with the people doing the job before designing the screens they will live in.
Every action is auditable and reversible, which is what makes people willing to use it.
Staged rollouts with a way back: any update can be halted and reversed from one place.
Power, connectivity and thermal limits are treated as requirements, because in the field they are.
[ SELECTED WORK · CLIENTS UNDER NDA ]
TAP FOR DETAIL
Eleven substation protocols normalised into one platform, 41k monitoring points. Control-room queries that took four minutes now return before the operator lets go of the mouse.
We wrote the on-vehicle software, the ingest and the planner interface across 2,300 vehicles. One team meant the arrival model could ask for a signal the hardware was not yet sending; 8% less fuel burned.
Listening devices, a detection model and a dispatch view for repair crews across 900km of mains. Time from leak to crew on site fell from days to hours, with every alert traceable to the sound behind it.
[ WHY CLIENTS CALL ]
the second attempt has to work
Most of this work arrives after a first attempt stalled: three suppliers, no single owner, and a deadline that has already moved once. We take the whole path and one accountable line.
How we work
WEEK 1
TEARDOWN
We read the firmware, the queries, the incident log and the tickets, then hand you a written map of what is actually running and where it leaks.
WEEK 2 — 3
thin slice
One reading travels from the device to a screen someone uses, live behind a flag. If the path cannot hold that, no roadmap will save it.
WEEK 4 — 12
BUILD OUT
Weekly deploys against a fixed scope. Tests, runbooks and dashboards land with the code, not in a later phase.
THE MODEL
set to suit you
Full handover, shared ownership or an ongoing managed service. We set that at the start and write it down, so nobody is surprised at the end.
[ THE LAB ]
BREADTH IS
PRACTISED, NOT
claimed
Two days a month are unbilled and spent on open tooling, benchmarks and write-ups. It is how a small team stays current across the whole path instead of expert in one part of it.
- arqen/streamcheck2.1k ★
- arqen/loadbench870 ★
- arqen/fleetwatch1.4k ★
Field notes, monthly. Numbers, failures and method.
Written by whoever was on call.
[ WHERE IT RUNS ]
11 INDUSTRIES
ENERGY & GRID
MONITORING · FORECASTING
logistics
ROUTING · FLEET OPERATIONS
MANUFACTURING
LINE CONTROL · QUALITY
public utilities
NETWORK MONITORING · BILLING
FINANCIAL DATA
TIME-CRITICAL DATA
healthtech
DEVICE AND RECORDS INTEGRATION
AGRITECH
REMOTE SENSING · YIELD
mobility
IN-VEHICLE · REMOTE UPDATES
Also telecoms, defence electronics and marine survey. Regulated, safety-critical and audited environments are the normal case here.
A substation, a factory line and a trading feed pose one problem in three dialects. Having shipped all three is why the second week is faster than the first.