I'm Kobe, a data engineer who loves computers and programming. I am always excited to analyse issues and dive into technical rabbit holes. I design and implement data solutions that directly benefit the business. As a cloud architect, I am proficient in developing cloud-based solutions that are both scalable and secure. If there is one thing I enjoy more than computers, it is working with people. I like to work in different teams or fields. When I am not working, I like to spend my time on passive investing, reading sci-fi books and volunteer work.
Lineas is the largest private rail freight operator in Europe. I develop and set up Kubernetes clusters, AWS data pools and NiFi data flows using Terraform and custom scripts, to provide a central data storage and management platform, while ensuring end-to-end functionality by corresponding with the BI team.
The ANPR Big Data Platform is a scalable and extensible platform for collecting, storing and aggregating ANPR data. My task within the project is to construct a big data lake using AWS and make the data usable for analysis, reporting and future processing.
ArtiFISHal Intelligence is a twofold project, primarily consisting of scenes made in Unity which illustrate several reinforcement learning techniques. Secondly, the project contains a simulation of a marine ecosystem which uses an open-ended algorithm, NEAT.
Sierteelt addresses a problem in floriculture: human influence & bias. An art exhibition was set up, with a Generative Adversarial Network (GAN) as a central piece. The AI generated plant-hand-like hybrids, which visualize human bias in plant structures.
The Covid-19 Dashboard consists of a full-fledged predictive AI, accessible through an administrator dashboard. The model was fed multiple Covid-19 data files, which trained it to predict future casualties.
The application allows users to explore local woods, nature & walking routes. An extensive Master/Detail Flow shows spots close to your location and the spot's description, reviews and pictures.
A website was built which asks the user how he/she feels. Upon entry, the website used a Natural Language Processing (NLP) model to extract the sentiment from the input and return a corresponding emoji.
Extensive setup of a multiple-container Docker environment with HDFS, NiFi, Kafka, Spark, ... Data flows in a recommendation system which used Spotify track data to recommend new fitting tracks.
The Turtlebot3 was simulated in Gazebo, using the ROS framework. The Turtlebot3 received positive rewards upon executing beneficial actions (using QLearn). The agent was eventually trained to race on a 3D-track.
The Clickbait Generator used metadata of viral Youtube videos to train a predictive AI model. The model learned to produce video titles and even short descriptions, based on a given category like 'Gaming' or 'Sports'.
Ducktrails is a Computer Vision application, based on the DuckieTown Platform. DuckieBots had to autonomously navigate through a grid, using a seperate webcam, pathfinding algorithms and an assistive GUI.
