Most companies describe research and development as a roadmap. We would rather describe it as a record. The code we have given away has reached more than a million developers, the platforms we sell were built and funded by us, the devices we design have been deployed for close to a decade, and the findings we think are worth sharing get published with the numbers attached.
The contributions MKN has made to open-source libraries and frameworks have reached over a million developers worldwide. No single client engagement has ever put our work in front of an audience that size. That work is public, it is attributable, and it is still running in codebases we will never see. Our answers on Stack Overflow are part of the same record. Contributing to the tools everyone depends on is the least glamorous form of research and the one with the longest half-life.
Our device work began with Espressif's original ESP chip, which made it cheap for the first time to put a radio and a microcontroller inside something small, and it has continued without a gap since. The output has been logistics devices, trackers, and controllers, taken through board layout, enclosure design, firmware, and volume production. Nearly all of it was built for clients and carries their names rather than ours. That decade of work is what the hardware engineering practice is built on, and MaverickTracker, launching soon, is the first device that will carry ours.
The IQ product family is funded and built by MKN rather than commissioned by a client. Each platform exists because we watched the same problem solved badly enough times to justify solving it properly: scheduling across divisions that refuse to share a calendar, tow and impound compliance that differs across all fifty states, permitting and licensing for local governments still working largely on paper. Building a product rather than a project imposes a discipline that client work does not, because there is nobody to hand the maintenance burden to when an engagement ends.
When we learn something we can prove, it goes in articles rather than a sales deck. The piece on how we reached a perfect Lighthouse score documents an actual measurement, including the part where our own analytics tag turned out to be the thing standing between us and the last five points. The piece on what actually makes a website slow ranks the causes by real impact and names the popular optimizations that are not worth anyone's time. Publishing the method next to the result is how we keep ourselves honest.
We are wary of research described only in the future tense. The items on this page point at things that exist and can be checked: a library you can read, a platform you can buy, a device that has been in the field for years, an article with its numbers shown. If you want to talk about work that does not exist yet, contact us and we will be straightforward about whether we are the right people to build it.