Ibrahim Beniffou
Description :
My project focused on the study of ammonia combustion using the RQL architecture in the SGT-800. Initially, we investigated an appropriate geometry for the primary zone, ensuring a self-sustaining flame, an efficient recirculation zone, and stable combustion. We provided best practices for designing this zone to achieve a favorable distribution of the quantities of interest. Subsequently, we shifted our attention to the aerodynamic study of the secondary zone, where we conducted optimization work on the injection angles to successfully maximize flow mixing at the outlet.
This experience has provided me with the opportunity to engage in a tangible industrial project, complete with well-defined goals and objectives that have practical applications in our daily lives. Throughout this internship, I have significantly enhanced both my communication and technical skills. Collaborating with professionals in a dynamic environment has allowed me to learn how to effectively tackle complex ideas and work as part of a team. Additionally, I have gained valuable insights into the technical aspects of combustion systems and the challenges associated with developing sustainable energy solutions. Overall, this internship has been instrumental in my professional growth and has equipped me with skills that will be beneficial in my future career.
Nur Cristian Sangiorgio
Description :
Richemont is a Swiss leading luxury goods group, committed to innovation and excellence. Its portfolio includes prestigious Maisons known for their craftsmanship and creativity (Jewellery, Specialist Watchmaking, Fashion & Accessories), alongside Online Distributors, employing over 40,000 people across more than 150 locations.
My internship was based within the Research & Innovation department at ValFleurier, one of Richemont’s factories dedicated to movements production, in Buttes, Switzerland.
During my internship, I focused on the implementation, testing, and validation of models and numerical methods designed to accurately reproduce the behavior of a mechanical watch movement. Initially, I dedicated time to gaining a comprehensive understanding of the operational principles of a watch movement and the underlying theory governing its subsystems. Subsequently, I translated this theoretical knowledge into the development of scientific code. Finally, I conducted numerical simulations, comparing the results against available experimental data to perform validation.
My work resulted in improved fidelity of watch movement numerical simulations. This advance is documented in an internal technical report, accompanied by a project presentation and the developed code. The dynamic and collaborative environment at Richemont contributed to the enhancement of both my modeling and simulation expertise, as well as my teamwork capabilities within a cross-functional and cutting-edge setting.
Figure: a mechanical watch movement
Maddalena Detti
Description :
Ezra Baup
Description :
MotionTech (Swiss Motion Technologies SA) is a medtech start-up launched in January 2017 and based in Lausanne. Its aim is to provide next generation mobility solutions to people living with disabilities. The first product, Your® Liner, is a custom-made prosthetic liner where each product needs to satisfy unique patient features; and it is the first 3D printed silicone liner. MotionTech is providing it to orthotic and prosthetic professionals throughout Europe, but also in Australia, New-Zealand and Canada.
The first step of the manufacturing operation consists of transforming a 3D scan of the limb into the 3D shape of the optimal prosthetic liner for the given user. To achieve this, the company has its own 3D toolkit, developed in C# with a user interface built using XAML Designer. The software is continuously updated to meet the needs of the designers, such as automating the modelling operations or adding new features to improve the liners quality.
My first task during the internship was to help enhancing this software by improving inspection analysis tools and optimizing some of the existing modelling operations – in terms of qualitative results on the liner’s shape and in terms of speed of use. In the second part of my internship, I developed a production planning tool to assist the manufacturing team in organizing their daily tasks, workload repartition, as well as estimating delivery times for new orders. This tool was developed in Python using constraint programming techniques.
Mathilde Simoni
Description :
LYO-X AG is a consulting company based in Basel, specializing in PK/PD modeling and Quantitative Systems Pharmacology. They assist pharmaceutical and biotech companies with modeling and simulation services throughout the drug development process, from target assessment to dose optimization. The focus is largely on biologics like monoclonal antibodies, as well as novel drug formats such as cell-based and gene therapies among others.
The objective of this internship was to investigate and model the distribution of monoclonal antibodies (mAbs) and antibody-drug conjugates (ADCs) in solid tumors, in order to better understand the key factors influencing tumor targeting and drug efficacy.
I began by working with the mechanistic compartmental model developed by Thurber et al. (2012), which describes antibody extravasation into tumors using a diffusion-based approach. I extended the model by incorporating additional biological mechanisms, including full antibody–receptor binding kinetics and intracellular label retention. The extended model demonstrated good agreement with experimental mouse data across multiple studies.
In the second part of the project, I developed a minimal physiologically based pharmacokinetic (mPBPK) model which describes the whole-body distribution of ADCs and their released payloads in humans. This model classifies healthy organs and tumors into compartments. It also includes mechanistic representations of the drug’s transport in plasma, lymph and tissues, receptor-mediated internalization, ADC’s deconjugation, as well as payload transport and clearance. It was calibrated using clinical data from two FDA-approved HER2-targeting ADCs and revealed differences in tumor toxin exposure between the two drugs, consistent with their differing mechanisms of action.
Figure: Extended mPBPK Model for ADCs
Source: Thurber, G. M., & Wittrup, K. D. (2012). A mechanistic compartmental model for total antibody uptake in tumors. Journal of theoretical biology, 314, 57-68.
Léa Gainon
Description :
The Kudelski Group is a leading Swiss company based in Cheseaux-sur-Lausanne (CH) and Phoenix (US), providing core digital security technologies and solutions for a wide range of customers in the fields of media, cybersecurity, and the Internet of Things (IoT). I joined the IoT Advanced Research Team for a six-month internship, focusing on the exploration of agricultural solutions using satellite imagery.
After an initial month dedicated to review existing work on precision agriculture and remote sensing, we decided to implement a tool to monitor crop growth using satellite data. The core of my internship consisted of building this project from scratch, ranging from data collection and scientific computations to the development of a user interface. I applied several mathematical algorithms from various fields, performed data analysis, and evaluated results.
Among the methods used were interpolation techniques, noise filtering, function fitting, trend detection, autoregressive and Bayesian models. This internship provided a great opportunity to apply the skills acquired during my master’s courses at EPFL in a real-world industrial context.
Last but not least, my team decided to initiate a patenting process for the idea I developed, which gave me valuable insight into intellectual property strategies and the steps involved in protecting innovation within a corporate setting.
Figure 1: User Interface for crop growth monitoring
Figure 2: Vegetation index map over Cheseaux from Sentinel 2 data. Red color refers to low index, green to
high.
Adrien Joliat
Description :
The European Space Agency (ESA) is Europe’s gateway to space. Founded in 1975, it is a 23-member intergovernmental organisation devoted to space exploration. My internship took place at ESTEC—the European Space Research and Technology Centre—in Noordwijk, the Netherlands. Within ESTEC, I worked on CHIME (Copernicus Hyperspectral Imaging Mission for the Environment), an Earth-observation mission planned around 2030 to advance food security, sustainable agriculture, and raw-materials monitoring.
During my internship, I developed an onboard AI model that detects marine anomalies over open ocean (e.g., floating plastics, oil spills, and sargassum). I carried the project end to end: dataset creation, model training, evaluation and validation, and implementation in CHIME’s dedicated simulator (CHEES). The outcome is a flight-practical linear SVM that can be used for onboard Marine Anomaly Detection within CHIME’s constraints.
Beyond the technical work, the CHIME team at ESTEC was the standout of my internship. Diverse in background yet perfectly aligned in mission, the group balanced scientific rigor with an inclusive, supportive atmosphere. Colleagues took time to follow my project, encouraged initiative, and celebrated small wins. Their guidance turned a demanding project into an energizing learning experience. In addition of this, meeting many other interns and young professionals is part of the ESA journey and adds real fun to the process. I highly recommend joining ESA for both the professional growth and the vibrant community it fosters.
Davor Dobrota
Description :
Julie Charlet
Presentation of the company :
Alpiq, headquartered in Lausanne, is a Swiss company specializing in electricity production and energy services. The company operates throughout Europe and has a workforce of around 1,200 employees. Its portfolio includes hydropower facilities as well as solar, wind and gas power plants. Company services are splitted into three main pillars: Assets, Trading and Origination.
Description of results :
The main outcome of this internship is a data analysis workflow and Python library designed for vibration monitoring of Pelton turbines during their runup phase. The library provides essential tools for:
– Reading and processing vibration data.
– Automatically analyzing and annotating runups in the time-frequency domain using the Short-Time Fourier Transform (see Figure 1).
– Generating ensemble averages to track the evolution of natural frequencies.
The results are visualized through a surveillance dashboard, which demonstrates how such a tool could be implemented for real-time or near-real-time monitoring during runup phases. This serves as a proof of concept for future operational monitoring solutions and lays the groundwork for predictive maintenance strategies that can minimize unplanned outages, extend turbine lifetime, and maximize asset availability.
Figure 1: Annotated runup sequence with fitted B-spline
Max Brodeur
Description :
My project at the Max Planck Institute for Brain Research in Frankfurt was focused on the connectomic analysis of the cuttlefish optic lobe – the entry point in the nervous system of a sophisticated camouflage system. My team’s aim was the reconstruction of a synaptic-level connectivity graph (the connectome) obtained from two volume Electron Microscopy (vEM) datasets of the optic lobe cortex. This is where the earliest steps of visual processing take place, similar to the vertebrate retina. To get the connectome, these datasets are first segmented into small fragments of cells that are iteratively merged to obtain reconstructions of entire neurons.
However, this automatic segment agglomeration process produces merger errors between different cells that drastically affect the quality of the connectome. My task was to train a contrastive learning model to predict segment cell types (e.g. Neurons vs. Glial cells) to prevent these common errors during reconstruction.
