INSA is a state engineering school, it is recognised for the excellence of its five-year education. I selected the IT department to get a overview of the different branches of software engineering.

In the Flight Test center I was in charge of the A350-900 certification for Laws and Handling Qualities. I joined Aerodynamics domain in 2016 where I'm working on HQ modeling, Performances, and Early design phases tools development. In this context I'm developping solutions based on Deep Learning.

In 2024, I founded TOST Unlimited a start-up dedicated to AI consulting and Custom Creation of Trusted Chatbots dedicated to e-commerce web sites. Our first customer is Konyks where you can find the chatbot integration here.

In 2018, I founded TOST Corp. a start-up dedicated to making and selling IoT devices such as Somfy shutters control hubs, Connected thermometers and QR Code as connected home ring.

I was the Technical Manager of a team of 15 people dedicated to writing more than 70 Certification Cards. I am familiar with CS-25 certification constraints and capable of devising certification strategies.

I'm a deep learning designer with coding skills using TensorFlow and PyTorch. I've used following techniques, MLP, CNN, GAN, LSTM, Graph Neural Networks,  Self Attention layers, PINNS. I mainly applied AI on aerodynamics data (Wind Tunnel, CFD, Flight Test). Good understanding of Autoregressive models, transformers, GPT,  embeddings and LLM in general.

I have completed management training courses and have experience leading teams in the development of software projects, CS-25 certification, and flight test analysis. 
Additionally, I have managed client relationships throughout my career.

Private Pilot Licence, In telemetry room for the A350-900 First Flight, Mont-Blanc summit, Marathon finisher, Presenter at the AIAA conference in Dallas.

Based on IoT temperature measurements, if there is a data gap due to battery depletion, an LSTM-based neural network can fill the gap in the time series. {Tensor Flow}

Use of a Graph Neural Network to predict pressure distribution on an unseen 2D airfoil profile. The GNN  captures the geometry of the profile and the associated pressures. The training dataset consists of various profiles within a Mach, Alpha, and Sweep domain. {PyTorch Geometric} 

As a 3D designer and printing specialist, I use a professional-grade 3D printer to bring ideas to life.  I have the ability to seamlessly transition from 3D printing prototypes to small-scale production. With my expertise in design and production, I'm able to produce high-quality, consistent results.