Arthur Gassner

I wrote a detailed write-up on the challenges one encounters when building ML solutions, from problem understanding to deployment. Both the frontend and backend are open-sourced.

I built a website challenging people to cycle across Switzerland -- end-to-end. Both the frontend and backend are open-sourced, with the backend hosted on a VPS.

I had the chance to present biped.ai at the Applied Machine Learning Days (AMLD) in front of more than 2000 people, as part of a startup spotlight led by Innovaud.

Using information extracted from the wireless signal of an LTE tower, one can try to estimate its position. This paper open-sources the first dataset of this kind using the Channel State Information (CSI) to do just that.

Created a data story exploiting bee-related datasets to better understand beekeeper migratory patterns and honey trades around the world.

We designed and built from scratch a 3D-printed test rig for multicopters, as well as the tuning software (based on ROS and PX4, written in Python)to assist the user in the tuning of the drone’s control parameters.

We developed effective distributed and reactive controllers for small groups of mobile robots with substantial hardware limitations.

We assessed various machine learning methods to predict whether a stock was undervalued or overvalued following an IPO.