Martin Mikušát

m.mikusat@gmail.com

Svelte or React on the web, Node on the server, Tensorflow for some light AI.

Focused on the simplest, lightning fast solutions to solve a particular problem.

Život v Trenčíne

98 Lighthouse score, 1.9s TTI

Made with Sapper

Website created to improve communication about issues related to life in my hometown. So far includes the following:

  • Landmark recognition

    Currently not live on the website as this requires more data to be useful which is hard to get to in the current situation. You can see a demo on my Instagram in the highlighted stories ("AI - pamiatky").

    All you have to do is point your phone camera at a landmark and an image recognition model will tell you which one it is. The whole model and inferrence happens locally on the user's device without any data leaving to server somewhere in order to preserve privacy and increase the speed on the whole process.

    For some, I also created 3D models using photogrammetry, which means that you would be able to play around with the landmark as well. I'm also working on a much more interactive version.

  • Correct recycling

    Every time I wanted to correctly recycling something, I had to decipher it from several documents which describe what belongs to which container. That doesn't solve my issue - I simply have a single thing I want to throw out, I don't need to know about everything.

    That's why I implemented a simple search where you can search for a particular thing and it tells you in which container that particular thing belongs.

  • Ecological alternatives

    Some things can be replaced by more ecological ones. For example, you can use wax napkins instead of celophane for food packaging. This is implemented as another search where you look for the thing you would like to replace and the results are any ecological alternatives.

    It's also integrated into the "Recycling" search. If you wanted to know where to throw away celophane it would not only tell you that but also what you could replace it with so that you don't have to produce any waste at all next time.

  • Local businesses

    The next step is to also get information on where to buy those ecological alternatives. You can find this information on the main page for each of the alternatives or you can, once again, simply search for what you would like to buy from local businesses in order to support them.

  • Trash collection schedules

    On the offical website, this information is hiden behind 5 webpages, in a PFD, in a complicated table you have to decipher. Instead of that, I formatted the data so that all you have to do is choose which city region you live in and immediately get just the relevant information.

  • Main search through the whole project

    Most experimental function. You can search for "Where to throw away celophane" or "How can I replace cotton swabs" and get the right information from the relevant section of the website. Implemented using (not so) clever formatting of the same data used elsewhere.

Warehouse

85 Lighthouse score, 3.3s TTI

Made with Gatsby. I'm currently rebuilding the site in Sapper.

Instant search through cameras, lenses, adapters, and memory cards using filters unavailable anywhere else. Includes a search through 85000 "kits" - valid combinations of a camera, lens, and optionaly an adapter.

Implemented by having all the data locally, loaded from a JSON file with a (trivial) custom compression. That way, the data has about 50kb, loads instantly on the page load, and allows for the instant search, even through the thousands of kits.

Also has a way to sort the results using arbitrary amount of parameters, e.g. if the minimum focal length is the same, use the maximum focal length, and so on.

Is it worth optimizing?

Miniature project that answers a single question - How long can you work on making a routine task more efficient before you're spending more time than you save?