Stephane d Ascoli
About
Stephane d Ascoli is from Greater Paris Metropolitan Region. Stephane is currently Research Scientist at Meta, located in Ville de Paris, Ile-de-France, France. In Stephane's previous role as a Ai4Science Research Fellow at Ecole polytechnique federale de Lausanne, Stephane worked in Lausanne, Vaud, Suisse until Oct 2023. Prior to joining Ecole polytechnique federale de Lausanne, Stephane was a Ph.D. Student at Ecole normale superieure and held the position of Ph.D. Student at Region de Paris, France. Prior to that, Stephane was a Ph.D. Student - FAIR at Facebook AI, based in Region de Paris, France from Sep 2019 to Feb 2023. Stephane started working as Author at Editions First in Aug 2019. From Sep 2019 to Mar 2020, Stephane was Teaching Assistant Professor at Ecole normale superieure, based in Region de Paris, France. Prior to that, Stephane was a Oral Examiner ('Colleur') at Ministere de l'Education nationale et de la Jeunesse, based in Region de Paris, France from Jan 2017 to Jan 2020. Stephane started working as Research Intern at Snips in Region de Paris, France in Mar 2019.
Stephane d Ascoli's contact information is available for free on finalscout.com, a web-based professional networking database with more than 500 million business contacts and 200 million company profiles.
Stephane d Ascoli's current jobs
Working in the Brain & AI team on decoding language from neural activity. https://ai.meta.com/blog/ai-speech-brain-activity/
Stephane d Ascoli's past jobs
Working with biologists, neuroscientists, chemists and physicists to leverage modern AI techniques. https://www.epfl.ch/research/domains/cis/epfl-ai4science/
Joint Ph.D. candidate between the Laboratory of Physics of Ecole Normale Superieure and Facebook AI Research. Main research topic : understanding and improving Transformers, with applications in Computer Vision and Symbolic Mathematics.
Joint Ph.D. candidate between the Laboratory of Physics of Ecole Normale Superieure and Facebook AI Research. Main research topic : understanding and improving deep neural networks.
Author of science books on AI and Physics for the general audience (see link below).
Teaching the graduate course on deep learning at ENS with Marc Lelarge.
Oral examinations in Physics for undergraduate students preparing in highly-selective institutions (lycees Henri-IV, Saint-Louis, Michelet) for the competitive entrance exams of the French 'Grandes Ecoles'.
Developed an open-source conditional variational auto-encoder for text generation, jointly supervised by Alice Coucke & Francesco Caltagirone (Snips) and Marc Lelarge (INRIA). Source code available here: https://github.com/snipsco/automatic-data-generation
4-year fellowship with a status of civil servant. Nationwide entry rank : 6th/1049.
Research at the Institut de Physique Theorique (IPhT) of CEA Saclay on the loss landscapes of deep neural networks. Supervised by Giulio Biroli, as member of the Simons Foundation collaboration on Cracking the Glass Problem.
During 6 months, I worked under supervision of Scott Noble (NASA Goddard Space Flight Center) and Manuela Campanelli (Center for Computational Relativity and Gravitation) on the simulation of the merger of a supermassive binary black hole. Running a general relativistic ray-tracing code on a supercomputer, we produced a highly realistic simulation of the electromagnetic counterparts to gravitational waves which may be observed in the next few years. Paper available here: https://arxiv.org/pdf/1806.05697 NASA press release: https://www.nasa.gov/feature/goddard/2018/new-simulation-sheds-light-on-spiraling-supermassive-black-holes APOD article: https://apod.nasa.gov/apod/ap181203.html
Cosmological simulations for the spectro-imager 'BATMAN'.