Avalor AI

Avalor AI, specializes in autonomous drone systems. Their drones are tailored for varied applications like forest fire monitoring, search and rescue operations, and battlefield mapping. During my time there, I was tasked with the development of an analytical trajectory generator using Rust. This project was split into two significant components.

Velocity Profile Generator

The algorithm focused on generating velocity profiles, my implementation used 15 sinusoidal segments, which were differentiable up and until the jerk. By constructing the profile using jerk continuous functions, we ensured a smooth trajectory with no spikes in the acceleration profile, which is pivotal for maintaining a consistent camera position during drone operations.

Multiwaypoint Navigation

The multi-waypoint navigation algorithm was designed to connect multiple waypoints. Using an analytical method, a path was drawn between these points. The integration of clothoids in the algorithm allowed the corners to be rounded in such a manner that the velocity profile of the drone did not need to be adjusted to incorporate sudden spikes. This was essential to prevent abrupt jumps in acceleration, ensuring the drone maintained a consistent flight pattern.

Combined, these algorithms formed a trajectory generator with minimized latency, wholly developed in Rust. My trajectory generator was subsequently integrated into Avalor AI's production systems.

NIBC N.V.

NIBC has provided me with the opportunity to participate in their innovation team. The goal of this team was to full fill the need of the employees for specialized software, which would enable them to enhance their performance and make more educated decisions. During my internship, two software implementations were created. Besides creating software implementations, the team made sure NIBC was up-to-date with the current developments in the tech word, and organized multiple events including hackathons for employees to participate in.

Mortgage Co2 footprint analyser

The carbon footprint of NIBC's mortage portfolio needed to be mapped and wrapped into a tool. This tool would help investors gain insights in the "greenness" of their current portfolio, and provide information about which homes could be bought to improve the score of their portfolio. To create such a tool, Power Bi was used, in combination with the coupling and data cleaning of multiple data sources which contained information regarding the energy label of houses. Predictions were made for incomplete data points using ML, which resulted in a total coverage of 89%.

The tool visually conveyed information to it's users using scatter plots layed on top of the map of the Netherlands, the charts were interactive, and with minimal user input, one could see all the statistics that mattered.

Natural language processing of news articles

News articles had to be constantly monitored, so that NIBC was update with all the semantics and market trends at all times, Employees had to manually scan which articles were interesting, of which the majority contained no important information at all. Analysing critical articles with employees, and extracting essential keywords and the polarity of these articles, a simple algorithm was developed which parsed the RSS news feeds through a python script, which was able to flag new articles if they contained the correct keywords, and scored neutral enough on the polarity scale. The tool was able to decrease the workload of employees, and implemented before the rise of ChatGPT. 

Eco-Movement

Eco-Movement is the leading provider for data on electric vehicle charging stations. As a fully independent data platform, they aggregate relevant information of (semi-)public charging networks all over the world into a uniform, accurate, and complete database.

During my internship at Eco-Movement, I worked as a data analyst. As a data analyst, I was responsible for collecting and organizing large datasets, generating queries which contained outliers, and ensure that these outliers were correctly supplemented with a team of three members.

In addition to data management and analysis, I played a key role in market research initiatives. This involved conducting surveys, analysing market trends and competitor behaviour, gathering insights on customer preference. My research findings helped the organization better understand their target audience and make informed business decisions.

To accomplish my tasks, I used a variety of tools and technologies, such as SQL for data management and querying, Excel for data analysis and visualization, and market research software for survey design and data collection.