I am a Data Scientist who specializes in data analysis and modeling utilizing scientific libraries, artificial intelligence frameworks, and the UNIX shell. Both Academia and Industry experience.

About Me

    🧠 With a broad scientific background in computer science and machine learning, I have gained invaluable experience through my diverse research and work experiences. Working as a PhD researcher in Group Neurophysiology at KU Leuven for 2 years, I conducted extensive research on deep Convolutional Neural Networks and their similarities with the visual system. Then, I transitioned to the role of Data Scientist at Faktion, a startup in Antwerp. During my tenure, I gained proficiency in various industry practices and applications, including setting up end-to-end ML pipelines, training models for AI tasks, creating deployable Docker containers, and working with Cloud components. I also participated in two hackathons and was proud to win one of them. 📊

    🧑‍🎓 I returned to KU Leuven to pursue applied research and obtain my PhD in Bioscience Engineering. Currently, as a Post-doctoral researcher at MeBioS, KU Leuven, I am working on building intelligent automated insect-monitoring systems that require high real-world performance based on data gathered by optical sensors, utilizing Machine Learning. Additionally, I am conducting research on hyperspectral imaging and supervising the theses of doctoral and master's level students. My role in each project spans multiple phases, from data collection and annotation to data analysis and model development. 🧑‍💻

    💻 My future plans include gaining a deeper understanding of MLOps tools and techniques to deliver my trained models to the cloud or embed them in field agricultural devices, such as insect traps or imaging setups. I have already developed a Streamlit app for insect classification and annotation tasks to aid researchers with their workflow. At the moment I am building a FastAPI server to connect our deployed model with various partners. Furthermore, I am learning Hyperspectral Imaging techniques to model multi-dimensional data more efficiently, enabling me to take on even more complex projects in the future. 📈


Here you can find a selection of personal projects I've worked on in the past, during my PhD:

  • PhotoBox: Software for imaging insect sticky traps and classifying detected insects.
  • stickybugs-ai: Insect image data analysis & modelling for classification/recognition purposes.
  • Home Surveillance: Motion detection using OpenCV (Raspberry Pi compatible), alerting through pushbullet, served with flask.
  • DL Tutorial: Tutorial for MSc. students on getting started with DL and image classification.
  • Flying insect trap: GUI for a Raspberry Pi connected to an optical Wingbeat Sensor and a Camera.
  • Wingbeat Frequencies: Messy code for the article: "Towards in-field insect monitoring..".
  • Knowing-Neurons article: A short article I wrote about Visual neurons and Deep Neural Netwokrs.



Knowing Neurons article

Home Surveillance

Deep Learning tutorial for BSc. and MSc. students (KUL)

Flying insect trap (private repo for now)

Wingbeat Frequencies


Slides from either seminars or group meetings.

Wingbeat signals

Insect Images

Intro to ANNs (3Blue1Brown, B.Rohrer)

PhD Defense


Feel free to send an email for any inquiries: kalfasyan [at] gmail [dot] com
Or reach me at any social media. Links provided at the bottom left of this page.