Independent IoT engineering
Measure it properly.Own what you build.
I design and build IoT systems end to end, from the sensor on the wall to the screen somebody checks every morning. You get the source code, the documentation and a team trained to run it without me.
No reseller commission
No per-device fees
Source code included
Worldwide
What I build
Sensor to screen, or the one piece you are missing
Some projects need the whole chain designed together. Others are three quarters built and need one part done properly. Both are ordinary work here.
Deciding what to build
Most of what goes wrong on an IoT project is decided in the first fortnight, by somebody choosing a radio before anyone has measured the site. I start with a coverage survey and a five-year cost comparison, both with your numbers in them, and the answer is regularly a hybrid that nobody proposed or a wired run that nobody wanted to hear about.
Choosing hardware that survives
Thousands of sensors measure the same handful of quantities and a good proportion of them fail in ways no datasheet mentions: cells that lose half their capacity below freezing, firmware with no back-off that flattens a battery retrying, gaskets that give up in their second winter. I shortlist, then field-test on your worst site with the failures induced deliberately, before anyone signs a purchase order for five hundred.
Devices and firmware
When nothing on the market fits, I build it: schematic, board, enclosure and firmware that holds up outdoors through a winter. More often the work is a bridge rather than a device, reading a machine that has no documentation left, and every file it took is yours at the end.
Water and leak detection
The measurement that finds a leak is not the total on the bill, it is the flow at four in the morning, when the honest answer should be zero. Clamp-on ultrasonic meters go on in twenty minutes without cutting pipe or dropping the supply, and a fifty litre an hour leak shows up the same night instead of on the next quarterly reading.
Energy and sub-metering
Metering per unit, per circuit or per machine, using clamp-on sensors that fit around conductors already in place so the switchboard stays as it is. The load profile that comes out has a habit of changing a capacity decision before it changes an energy bill, which is where the money usually turns out to be.
Sites with no power and no network
Weather masts, pumping stations, river monitors, tanks in a field. The sensor is the easy part; what decides whether it still reports in five years is a panel sized against the darkest fortnight of the year rather than the annual average, a cell that holds capacity below freezing, and firmware that buffers a backlog through an outage instead of losing it.
Equipment that predates IoT
A meter from the nineties and a machine with fifteen years left in it are not replaced so that somebody can have a dashboard. Nearly all of it already speaks: a pulse per litre, a Modbus register, a current loop, a fault relay. The work is reverse engineering the register map with the machine in front of me and listening without ever gaining a write path.
The data layer underneath
Most systems I inherit contain a small service nobody planned, gluing readings in one database to device records in another so a dashboard can draw. It is where the bugs live and it never gets deleted. Put both halves in one place and the join is just a join, which is the difference between a new report taking an hour and taking three weeks.
Dashboards people actually open
Fifty panels covering every available metric is an inventory of the database, not a dashboard, and nobody opens it twice. The ones that get checked daily have six or eight panels and each answers a question somebody genuinely has at nine in the morning. Ask mine for a full year and it draws it, without a spinner. There is one running on real devices if you want to see.
Where projects go wrong
The expensive mistakes are all made early
Almost every overrun I am called in to fix traces back to a decision taken before anyone had measured anything. The technology worked in the lab. The bill arrives at scale.
Year one
The radio chosen in a meeting
Somebody arrives with a technology already picked, and the rest of the project is spent making the site fit it. The three questions that actually decide it are where the power comes from, who owns the ground the devices sit on, and what the connectivity line costs in year five. All three are answerable in an afternoon, and answering them later means buying twice.
Year two
The pilot that proved nothing
Twenty sensors in one accessible building, maintained daily by the person who wanted them, with no security review and a spreadsheet standing in for a pipeline. Everything that made it succeed is what stops it scaling, and a pilot that surfaces no problems has not tested anything. Put it on the awkward site instead.
Year three
The subscription nobody modelled
Per-device fees are small on the quotation and larger than the entire hardware budget by the second year, which is when growing the deployment starts feeling expensive for no engineering reason. The five-year arithmetic is worth doing before committing, and sometimes it says to pay the subscription. It should say so with numbers attached.
A dashboard that answers instantly
Meters, air quality sensors, doors and a weather station, all reporting right now on real hardware. Drag across any chart to zoom, or ask it for a full year. It just draws it, with no spinner and no loading screen apologising at you.
Open the live dashboard
What you keep
The system is yours when I leave
Not a licence, not a tenancy, not an account that expires. The repository, the documentation and a team that can operate it without calling me.
01
You hold the code
Every engagement ends with the repository, the schematics where there is hardware, the register maps and calibration notes where there is a retrofit, and documentation written for whoever inherits it rather than for me. Bring in any developer to extend it. Nothing about that is conditional on our relationship continuing.
02
You hold the keys
Root keys, session credentials and decrypted readings stay on infrastructure you chose, which is what lets you answer your security team in one sentence. Retention, access and data residency are settings in your console rather than someone else's. That distinction is what gets a rollout through review.
03
Growth stays cheap
The two thousandth device costs the price of the device. There is no metered tier, no seat count and no renewal conversation, which matters more than it sounds: scope on these projects always grows, and it is only painful when every addition is another line in the operating budget forever.
Working together
An hour of advice, or the whole thing built
I have no reseller agreements and take no commission from any manufacturer, which is the only reason a recommendation from me is worth reading. If your problem is better solved by something I do not sell, that is what I will tell you.
Review
A second opinion
I go through what you already have and say where it will bottleneck, which is usually the placement of the field hardware, the shape of the ingest path, or a storage decision made in week two that nobody has revisited. A single session is often enough, and I would rather tell you the architecture is fine than find work in it.
Delivery
Built end to end
Devices, connectivity, ingest, storage and the screens above them, designed together rather than assembled from five parts that were each specified in isolation. The result feeds whatever you already run, over MQTT, HTTP, a queue or a direct database connection, because the job is to fit your operation rather than to become it.
Handover
Trained to run it
Sessions are hands-on and taught against your real system rather than a demo, covering provisioning, administration, the data pipeline and dashboard work. The goal is a team that does not need me, which is a strange thing to sell and the only honest way to leave a deployment behind.
Writing
Read all articles
Written from systems that are running
The measurements, the failure modes and the costs, left in. I only write about what I have actually built and measured.
Tell me what you are building
The first conversation costs nothing and usually saves more than it should. You will get a straight read on what the project takes, roughly what it costs, and whether the approach you have in mind is the right one. If I am not the right person for it, I will say so.
Get in touch