With the rise of IoT, GPS tracking is more available. Take for example a transport vehicle moving some materials. The geolocation data can tell us more than just where the vehicle is. Using geospatial analytics we can observe the mobility patterns of the vehicle. Clustering the data results in places where the vehicle stopped for work or parking. Geospatial analytics can tell us if the vehicle is working, find the optimal route, spot any unusual behaviour of the vehicle (e.g. unexpected stop), what materials are where and much more.

Bill of Material (BOM) is a highly used data representation in the Industry domain. It is used in engineering for both design and production processes. Each new or updated model of the machine means changes to the BOM, which can introduce human errors. For example, you want to upgrade the engine of a car, but overlook the limitations of the car suspension that might be incompatible with the new engine. In this project, we developed a way to automatically search for such mistakes using Machine Learning.

The business idea of identification of sensitive data needs no introduction. In my opinion, it is one of the most popular use-cases for Machine Learning. In this simple application (github.com/konvica/read-presonal-data), I leverage several libraries (spacy, presidio, pymupdf, pytesseract, OpenCV, streamlit) to process PDF or image documents and extract all personal and sensitive data like photos of a face or personal information.

In this minimalistic demo (github.com/konvica/spot-img-diff), I demonstrated a simple method to localize differences between misaligned images. For example highlighting differences between screenshots of a webpage, which can be useful for Web UI Testing.

Proof of concept implementation of Weather Chatbot using Rasa NLU and tommorow.io weather API. The chatbot looks for location in user input and provides real-time weather reports for a given location. Source files are in the repository github.com/konvica/mini-weather-chatbot and the whole service can be deployed using docker-compose and free tommorow.io API token.

Please consider looking at my CV here. You can see a brief summary of all projects I worked on in the About section. Feel free to write me an email or message me on Linkedin.