Magister is proud to present our 15 ESRs who will be working on the project in the coming years:

Click on the ESR to find out more about them and their research. (More to come!)

ESR 1: Ushnish Sengupta (UCAM)

Academic Division: Information Engineering

Research group: Computational and Biological Learning


Personal website

Research interests

Broadly speaking, I am interested in high-stakes applications of probabilistic machine learning techniques. The consequences of failure for an ML algorithm that monitors the sensors of an aircraft engine are very different from one that recognizes faces in social media photos or recommends music. These critical applications often place a high premium on principled uncertainty estimates,  applicability to limited datasets and interpretability: something probabilistic machine learning techniques like Gaussian processes can offer. Marrying these completely data-driven techniques with physical modeling for more robust predictions and sensible extrapolations is also something that intrigues me.

I am a Marie Sklodowska-Curie Early Stage Researcher in the MAGISTER consortium which seeks to utilize machine learning to understand and predict thermoacoustic oscillations in aircraft engines or gas turbines. My job, as I see it, is to serve as a liaison between the probabilistic machine learning group led by my PhD supervisor Professor Carl Rasmussen and the flow instability and adjoint optimization group led by my advisor Professor Matthew Juniper. We are currently looking at data from both small-scale and large-scale combustors to explore how ML techniques can use this data to enable both better designs and safe operation for these machines.

I am also a dilettante computational chemist and am curious about how probabilistic machine learning can improve our ability to predict protein aggregation and protein dynamics. Protein aggregates, of course, play both functional and pathological roles in the human body while protein dynamics is crucial to the functioning of many enzymes. Compared to the static structure prediction problem, however, both aggregation and dynamics are harder to characterize experimentally and lack extensive databases. Can Bayesian techniques shine in this data-limited regime and achieve results comparable to expensive simulations which consume many thousand of supercomputer core-hours?


I did my bachelor in Mechanical Engineering from the Indian Institute of Technology, Kharagpur and my masters in computational science from RWTH Aachen University, Germany. As a bachelor student, I worked on the computational modeling of compartment fires and microcombustors. In my masters thesis work, on the other hand, I analyzed data from molecular dynamics simulations and focused on the automatic generation of hidden Markov models to help computational scientists effortlessly derive a simple, concise "states and rates" picture from the massive amount of data they generate. My experiences shaped me into a person with great passion for both mechanical engineering and data science and I believe that the topic of my PhD represents a perfect fusion of these interests.

ESR 2: Nils Wilhelmsen (Armines)

Introduction and contact info:


Name: Nils Wilhelmsen

Work address: Centre Automatique et Systèmes

                        MINES ParisTech

                        60 Boulevard Saint-Michel

                        75006 Paris




Project partner: ARMINES(Association pour la recherche et le développement des méthodes et processus industriels)

Project title: Supervised Learning Algorithms for Distributed Parameter Models of Thermoacoustic Oscillations

Previous background:

MSc Engineering Cybernetics from NTNU(Norwegian University of Science and Technology), June 2018

Title of MSc Thesis: Minimum Time Bilateral Observer Design for 2X2 Systems of Linear Hyperbolic PDEs - With Application to Oil Well Drilling State Estimation for Improved Kick Handling


Courses taken in relation to project:

  • Fundamentals of thermoacoustic instabilities, CERFACS, July 2018
  • Probabilistic Machine Learning, MAGISTER Workshop, September 2018
  • Thermoacoustics and Combustion Dynamics, MAGISTER Summer School, September 2018
  • Flatness Based Nonlinear Control, MINES ParisTech, March 2019
  • Numerical Methods for Large Eddy Simulations, CERFACS, April 2019
ESR 3: Nilam Tathawadekar (GE)

My current research focuses on machine learning in the area of combustion instabilities. The basic idea is to improve sub-models of premixed combustor using deep learning techniques. Since September 2018, I am an Early Stage Researcher in General Electric (GE) Aviation Digital, Munich.

I am alumni of Indian Institute of Science, Bangalore. Here is the link to my work on eRetail demand forecasting using deep neural networks.

Contact :

E-Mail :

LinkedIn : Nilam Tathawadekar

ESR 4: Louise Da Costa Ramos (ANSYS)

Louise da Costa Ramos 




Project: Marie Curie Early Stage Researcher (ESR)

Academic host: Mines Paristech 

Industrial host: ANSYS

Supervisors: Florent Di Meglio (Mines Paristech), Valéry Morgenthaler (ANSYS), Luís Fernando Silva (PUC-Rio).

Topic: Development of a computational model using Deep Learning approach to enable combustion/acoustic coupling.

Software used: Fluent, ANSYS Tools, Matlab.



  • MSC Mechanical Engineering from Pontíficia Universidade Católica do Rio de Janeiro (PUC-RIO).

Project: Computational modeling of laminar non-premixed diffusion flames with detailed kinetic mechanism, by using OpenFOAM.

Supervisor: Luís Fernando Figueira da Silva - PUC-Rio

ESR 5: Sagar Kulkarni (TUM)

I am Sagar Kulkarni, a Marie Curie Early Stage Researcher at TU Munich. I am from India and have an M.Sc., in Mechanical Engineering from TU Delft, Netherlands. For my Master thesis, I carried out CFD analysis of a Micro Gas Turbine Flameless Oxidation spray burner for range extension application for electric vehicles at the Institute of Combustion Technology, DLR, Cologne and TU Delft.

As an ESR my work packages include LES of spray combustion for low order modelling of the dynamics and Uncertainty Quantification. In detail, I will be carrying out LES of spray flame in presence of acoustics in AVBP to determine the Flame Transfer Function (FTF) of the system. Then use the System Identification procedure to quantify the uncertainty of the FTF with respect to simulation parameters such as length of the time series and number of droplet parcels modelled.

As a result, I look at Flame Impulse Response (IR) and transfer function of the liquid fuel combustion, analyse the contribution of various physical processes (eg. atomization, evaporation) to the overall flame response in terms of frequency response functions.

I look forward to the cooperation from other partners (academic and industrial) towards the project to really deliver the goals and make a step towards mitigating instabilities from aero engines!!

ESR 8: Francesco Garita (UCAM)

Francesco Garita


Telephone: +44 1223 746971


Research interests

Project type: Marie Curie Early Stage Researcher (ESR) funded by the European Union
Topic: Physics-Based Machine Learning in Thermoacoustics
Description: Adjoint-based sensitivity analysis is combined with automated experiments so that a computer can create an accurate model of a combustion system and eliminate thermoacoustic oscillations, without knowing all model parameters a priori

Peer-Reviewed Publications

L. Marocco, F. Garita, Large Eddy Simulation of liquid metal turbulent mixed convection in a vertical concentric annulus. Journal of Heat Transfer, 140 (7), (2018).
doi: 10.1115/1.4038858


Ph.D. in Engineering

University of Cambridge, UK

     01/2018 - present

Internship for M.Sc. Thesis

Karlsruhe Institute of Technology, Germany

     09/2016 - 03/2017

Exchange M.Sc. Student in Mechanical Engineering

ETH Zurich, Switzerland

     09/2015 - 09/2016

M.Sc. (summa cum laude) in Energy Engineering

Politecnico di Milano, Italy

     10/2014 - 04/2017

B.Sc. (summa cum laude) in Energy Engineering

Politecnico di Milano, Italy

     09/2011 - 07/2014

ESR 9: Alireza Javareshkian (TUM)

Starting from the middle of March, 

Alireza Javareshkian 

started his project at the institute. The aim of this project which is funded by EU project-MAGISTER (H2020-MSCA-ITN-2017), is further investigation of acoustic behavior of combustor liners with dilution holes in aero GT engines. Characterization and modelling of acoustically absorbing liners would be performed through the course of this project, aiming acoustic characterization of perforated medium, measurement of damping rates of a combustion system with and without perforated medium and benchmark of measured and predicted (1D-network, LNSE) results.

ESR 10: Edmond Shehadi (UT)


After obtaining my B.E. in mechanical engineering at the Lebanese American University, Lebanon, I decided to proceed down the computational branch of science. Accordingly, I devoted the next couple of years of my career to earn my M.Sc. in computational science at Uppsala University, Sweden. There, I developed a great fascination towards computational fluid dynamics (CFD), especially turbulence with respect to large-eddy simulation (LES). Armed with the motivation, curiosity, and passion, this led me to become a PhD candidate at the University of Twente as part of a Marie Curie EU-funded project, MAGISTER.

My graduate school thesis tackled canonical flow cases in CFD, such as turbulent channel flows and backward-facing steps, via the open-source software, OpenFOAM and using wall-resolved LES. My M.Sc. work titled “Large Eddy Simulation of Turbulent Flow over a Backward-Facing Step” can be found here, while my past, present and future research interests and progress may be located on ResearchGate (see below).

Besides my professional career, some of my other hobbies include reading literature, playing and writing music and, for some weird reason, telling everyone that I know how to play tennis.

 Current Research

In the meantime, my work is heavily computationally oriented. It involves working with the relatively new open-source CFD code SU2, particularly the high-order discontinuous Galerkin (dG) solver therein. My current work tackles open-source development of non-reflecting boundary conditions, mainly via the perfectly matched layer (PML) technique and using wall-modeled (WM-)LES.

In the context of MAGISTER, my objective is to utilize the high-order dG solver of SU2, in conjunction with PML and WM-LES in order to better simulate (compressible) turbulent flow through combustor liners and dilution holes of jet engines – operating at transonic speed regimes. These, in turn, will further facilitate the understanding of the acoustics and turbulence mechanisms acting in such confined geometries by means of high-fidelity simulations. Concurrently, these will serve as additional training datasets for later machine learning purposes.


LinkedIn:        Edmond K. Shehadi

ResearchGate: Edmond K. Shehadi

ESR 11: Thomas Christou (KIT)

Born and grew up in Athens, Greece 


Studied Mechanical Engineering at the

National Technical University of Athens (N.T.U.A.)

5-year-Diploma (Dipl.-Ing.)

Specialization (4th and 5th year) on Energy and Process Engineering

Diploma thesis on optimization of different co-generation (electricity and cooling) configurations by utilizing a waste heat source

Internship for 2 months at Daikin Greece S.A. in Athens, part of Daikin Europe N.V., a serious air conditioning company based in Ostend, Belgium

Research start at the Karlsruhe Institute of Technology (K.I.T.)

in Karlsruhe, Germany

Department of Chemical and Process Engineering, Engler-Bunte-Institute (EBI), Chair of Combustion Technology

PhD Candidate and Early Stage Researcher (ESR) for the Marie Skłodowska-Curie Actions (MSCA) project MAGISTER

Research focus on atomization and sprays 

Experimental approach on air blast atomization under oscillating flow field, in order to determine the acoustic influence of the air on the generated spray

Build of atmospheric test rig from scratch, with atomization nozzles developed and manufactured at KIT and pulsation device from Technische Universität München (TUM).

Main measurement technique:

Phase Doppler Anemometry (PDA

ESR 12: Sara Navarro Arredondo (UT)

Project title: Characterization of acoustically (un)forced kerosene spray flames at elevated pressure and preheated air

Experimental investigation in liquid fuel combustion. An air blast type burner will be used in a combustor with variable outlet conditions. Combustor thermoacoustics will be characterized by mean of dynamic pressures and OH* chemiluminescence recording. With the data obtained, prediction of thermoacoustic instabilities with machine learning will be proposed.

Location: University of Twente 

Supervisor: J.B.W. Kok

Place of secondments:

Industrial: General Electric (Switzerland),
Academic: Karlsruher Institut für Technologie (Germany)


MSc Chemistry, mention Chemical engineering at the Université Pierre et Marie Curie, France. November 2017

MSc Thesis “A model of the disjoining pressure by Dissipative Particle Dynamics (DPD) method” at the Institute Français du Petrole, Energies Nouvelles, France.


ESR 14: Thomas Lafarge (Safran Tech)

ESR 14: Thomas LAFARGE 


Master in engineering, Ecole Centrale de Lyon (France): Generalist engineer

Master in Science and Technology, Keio University (Japan): School of Science for Open and Environmental Systems.

Host entity: Safran Tech (Magny-les-Hameaux, France)

Academic host: CERFACS (Toulouse, France)

Research Topic: Investigation of the use of Lattice-Boltzmann method applied to multiphasic flows:

The numerical study of injection is a critical point if we consider the behaviour of a flame as both the flame and the injection responses to thermo-acoustic waves are coupled. Consequently, the study of the physic of injection is sensitive if we want to determine the physic of a combustion chamber.

A lot of current numerical simulation of multiphasic flows relies on front tracking methods, that are computationally expensive and not always efficient to catch breaking phenomena and particularly primary atomization. On another hand, Lattice-Boltzmann (LB) methods have recently shown a great degree of maturity and could be able to simulate multiphasic problems with high-density ratio in a close future. The first part of this work consists in investigating the potential of those methods on our applications. Afterwards, we will consider the use of machine learning in determining automatically the tunable parameters that are inherent to LB methods.

ESR 15: Pasquale Walter Agostinelli (Safran HE)


Pasquale Walter Agostinelli

Born in Benevento, Italy on the 1st of September 1994


Host institution: Safran Helicopter Engines, Bordes, France

Academic institution: CERFACS, Toulouse, France

University: Institut national polytechnique de Toulouse, Toulouse, France

Ciao! I’m Walter Agostinelli and I am an Early-Stage Researcher for Safran Helicopter Engines in the MAGISTER Marie Sklodowska Curie ITN European project. I have a BSc in Aerospace Engineering at Università di Napoli Federico II with a bachelor thesis at the Italian Aerospace Research Center, where I worked on the aerothermodynamic analyses of the GHIBLI Plasma Wind Tunnel. In July 2018, I obtained the Space Engineering MSc from Politecnico di Milano, the Double MSc Degree in Aerospace Engineering from Politecnico di Torino and in February 2019 I have received also the Diploma from Alta Scuola Politecnica, a management and business school funded by Politecnico di Milano and Politecnico di Torino. During my studies, after an internship at the University of Texas at Arlington thanks to a scholarship of the Italian Space Agency, I joined the Erasmus Program at TU Delft, where I was also Teaching Assistant for the course of Dynamics. I worked on my Master Thesis at the Von Karman Institute for Fluid Dynamics on the experimental and numerical characterization of the H3 hypersonic wind tunnel. I’m found of research and innovation and I like to work in a dynamic, stimulating and international environment.





Ushnish Sengupta

University of Cambridge


Nils Wilhelmsen


Nilam Tathawadekar



Louise da Costa Ramos



Sagar Kulkarni

Technische Universität München


Varun Shastry



Alireza Ghasemi Khourinia

University of Twente


Francesco Garita

University of Cambridge


Alireza Javareshkian

Technische Universität München


Edmond Shehadi

University of Twente


Thomas Christou

Karlsruher Institute für Technologie


Sara Navarro Arredondo

University of Twente


Michael McCartney



Thomas Lafarge



Pasquale Agostinelli 

Safran HE