Computer scientist and researcher at the Università di Genova.
I'm Arnaud Ruymaekers. I am a researcher in computer science at the Università di Genova. I currently work within the PiMLB unit (Physics informed Machine Learning for biological Behavior) of the Malga lab (Machine Learning Genova Center).
My research focusses on model-based reinforcement learning to the field of olfactory navigation in turbulent environments. I mainly work on using the concept of Partially Observable Markov Decision Processes. I work on finding solutions to simulate the behaviors of animals in order to undertand better how their sense of smell work.
I finished a Master's Degree at the Università di Genova in March of 2024 after completing my thesis on olfactory navigation. Before that, I took part in the Data Science and Knowledge Engineering degree at Maastricht University. During that degree, I also took part in the Honour's program Ke@Work which allowed me to gain some experience working for a company, Open Line. I spent 2 years as an intern over there and completed my Bachelor's thesis there. Following this, I continued working as a consultant for the company for the following three years focussing on Business Intelligence as a Data Engineer.
Overall, I really like programming, and am curious about many topics computer science related. During my Bachelor's degree I mainly used Java as a programming language. But following this, during my internship at Open Line, I got my hands on Python and database manipulation with SQL. I also learned to handle big amounts of data on daily using the ETL tools provided within the Microsoft Azure suite. Namely I worked using the Databricks platform where I got to work with the PySpark toolbox. Within my time at openline, I also learned a great deal about Source Control and development cycles. I got to help build a strong DevOps release pipeline that aimed at maximizing the data-availability for the company while allowing the Business Intelligence team of developping with as much peace of mind as possible.
I work within the PiMLB unit (Physics informed Machine Learning for biological Behavior) of Malga (Machine Learning Genova Center). Within this group, I work on olfactory navigation and, more particularly, model-based algorithms. These algorithms attempt to find optimal strategies for agents to find the source of an odor.
I helped maintain and further develop the Business Intelligence (BI) infrastructure of the company using Microsoft Azure services. The BI team was tasked with the Extract, Transform, and Load (ETL) process of the data of the company. The ETL process was built using Databricks, DataFactory and the SSIS package of Microsoft Visual Studio.
I also built software frameworks to speed up the processing of data to the various databases. This framework was also meant to simplify the daily operations of the BI team.
I also managed and improved the CICD pipeline within DevOps of the team to ensure optimal and smooth deployment of new features.
As part of the Honours program of the Data Science and Knowledge Engineering Bachelor of Maastricht University I integrated the Business Intelligence (BI) team of Open Line. During my internship I undertook projects related to Data Science and Optimization.
I developed an model for Time-Series forecasting applied to the forecasting of the amount of calls the various teams would receive in given periods.
My largest project there, which was also my Bachelor's thesis, was an Operation Research problem. I got to develop an Integer Linear Programming (ILP) solution to suggest improvements within the virtualization of storage. These suggestions would be monitored regularly to apply the changes. The suggestions provided by the algorithm allowed for a considerable amount of money to be saved on a monthly basis.
The solutions that I proposed also had to be efficient enough to be run on a daily basis as part of the morning Extract, Transform, and Load (ETL) process. Along this, they were included within this process in order to benefit from fresh data every data and to provide the most up-to-date suggestions ad predictions.
As the treasurer for the student association for two years (board 25 and 26), I succesfully managed the money flows of the organization while allowing very succesfull events to happen smoothly. Doing so while holding a balance of trying to keep the prices for the students during events as low as possible as the organization is not for profit, while also keeping enough to be able to plan ever more exciting events.
During my time at MSV Incognito, I also lead the yearbook commity which involved leading a group tasked to take pictures during events and write entertaining summaries to fill the yearbook with everlasting memories for all the students.
As a student ambassador for the Maastricht University I promoted my study program (Data Science and Knowledge Engineering) and the university to prospective students. As part of this, I helped during Open Days where I told my experience as a student and tutored a programming exercise meant for high-school students. I also guided prospective students taking part in the activity "Student for a Day". During this activity, I guided them through a day at university letting them experience what they could potentially be part of in the following years.
Track: Data Science & Engineering - Artificial Intelligence
During my Master's degree, I deepened my knowledge in the fields of data science and machine learning. I got to learn about topics such as Computer Vision, Natural Language Processing, Speach Processing and Recognition, Multi-Agent Systems, or Computer Games. The various project I worked on for these various classes can be found on the Project Page.
I joined the PiMLB unit (Physics informed Machine Learning for biological Behavior) of the Malga (Machine Learning Genova Center) lab to work on my Master's thesis.
Thesis Title: "Navigating by Scent: A Model-Based Approach to Olfactory Navigation using Partially Observable Markov Decision Processes" (Supervisors: Prof. A. Seminara; Prof. A. Verri) (Open in new tab )
Short Summary: A solution to the Olfactory Navigation problem using a model-based Reinforcement Learning framework: Partially-Observable Markov Decision Processes (POMDPs). A multithreaded, GPU-capable implementation of the Point-Based Value Iteration (PBVI) solver was implemented leading us to confirm the apparition of casting and surging behaviors. Some potential shortcomings of the technique where also uncovered.
Final Grade: 110/110 (cum Laude)
During my bachelor, I learned the fundamental of Computer Science and applied Mathematics to this domain. I also got to learn the fields of Data Science and Machine Learning.
The programming language I learned within this degree were: Java, SQL, Matlab, and Python.
I also took part in the Honour's program Ke@Work that lasted for two year and where I was placed at a company (Open Line) as an intern. My Bachelor Thesis was also written as part of this program and applied to the needs of the company.
Thesis Title: "Pareto Optimal State Search Using Simulated Annealing" (Supervisor: Dr. G. Stamoulis)
Short Summary: An Operation Research problem solved by Simulated Annealing and Integer Linear Programming.
Final Grade: 8.43/10
Exchange student for a year with the organization WEP. I took part of the Senior year in High School to learn to become fluent in English in order to pursue my studies at the Maastricht University.
During this school year, I took classes of AP Calculus, Web Design, among other.
I also took part in the Robotics Club and the Swimming team.
Option: Science with advanced Math (8h/week)
During my time in High School, I took part in the Science track with an advanced Math placement. Along this, I was also part of an advanced Dutch-language program with which I had more classes in Dutch than the regular program.