.
      I do this:
    Our project for helping children learn vector addition in a playful way
                
                2017-2021   
              
                      
                         
                      
                     | 
                    
                      
                         
                      
                     | 
                    
                      
                         
                      
                     | 
                  
| The red dots on paper are recognized by the qml-ar software (in blue) to know where the phone is. The chest is an augmented reality object. | Architecture of the qml-ar library showing the data flow. Each component is a C++ class, a QML file, or a physical device. | Test run of the learning activity at the International School of Geneva in 2019. | 
              
              
              Master's Thesis and a Google internship: simple world models lead to good abstractions
                
                2019-2021   
              
                      
                         
                      
                      Google Internship project: Our method obtains a causal model of an RL environment  | 
                    
                      
                         
                      
                      Thesis project: learning a simple causal model with interpretable nodes from pixels  | 
                  
                      
                         
                      
                      Process of learning the causal graph in the thesis project  | 
                    
                      
                         
                      
                      λ can be seen as annealing temperature, and non-zero components as -quality of the alloy.  | 
                  
DIY Quadcopter controller with an IMU and PIDs on Arduino Due
                
                2011-2014   
              
Tournesol: byzantine-resilient contributor-driven open-source open-data non-commercial recommender
                engine
                (TLDR: contributors compare pairs of videos and the website aggregates the ratings)
                
                
                2020-2021   
              
              Defending against adversarial policies in Reinforcement Learning via alternating training
                
                2020   
              
One complex escape room hardware/software & One b2b project for escape rooms
                
                2014-2016   
              
| Other projects | ||
| 🔎🦄 | 
              I am looking for
               | 
          
| ❔💬 | 
               
                I am interested in creative projects involving  
                My GitHub is  
                I  
                I am originally from   |