Resume

🎓 Education

Ph.D. - Signal Processing, Artificial Intelligence & Video Compression Jun 2023 – Jun 2026
Université de Rennes · INRIA team COMPACT · MediaKind (CIFRE)
Thesis: "Learned video downscaling for end-to-end Rate-Distortion optimization of video streaming systems"
Industrial supervisors: J. Le Tanou, M. Ropert (MediaKind)
Academic supervisors: T. Maugey, A. Roumy (INRIA)
Deep Learning Video Compression Rate-Distortion Optimization Surrogate Gradients ABR Streaming
M.Ing. - Electronic & Computer Engineering 2020 – 2023
INSA Rennes - Institut National des Sciences Appliquées

Embedded systems design (hardware & software), digital/analog signal processing, image analysis, AI (NN, CNN, transfer learning, GANs), VHDL, real-time systems.

C/C++ Signal Processing Deep learning
M.Eng. - Aerospace & Aeronautical Engineering (Exchange semester) Sep 2022 – Jan 2023
École de technologie supérieure (ÉTS) · Montréal, QC, Canada

Selected courses: fly-by-wire systems (MGA804), video communication systems (MTI810), DSP architecture (SYS835), introduction to avionics (GPA745).

C++ DSP Avionics
CUPGE - Computer Science & Engineering (Preparatory cycle) 2018 – 2020
ESIR - École Supérieure d'Ingénieurs de Rennes

Rank: 4th / 45. Foundations of IT, functional and OO programming, digital/analog electronics, mathematical engineering.

💼 Experience

PhD Intern - Deep Learning for Video Compression Nov 2026 – Feb 2027
Disney Research|Studios · Zurich, Switzerland
  • Research internship on deep learning methods for video compression.
Deep Learning Video Compression
PhD Researcher - Deep Learning for Video Compression Aug 2023 – present
MediaKind · Rennes, France (Hybrid) - CIFRE Partnership with INRIA
  • Developed surrogate gradient methods enabling end-to-end training through non-differentiable codecs (H.264/AVC, H.266/VVC).
  • Proposed a geometric interpretation of the SCALED surrogate gradient as an oblique projection, establishing its MSE-optimality under intensity-shift invariance.
  • Extended the framework to full neural wrapper optimization (pre- and post-processing), achieving BD-Rate (PSNR) reductions up to −23.59% on x264 and −20.07% on VVenC.
  • Paper accepted at PCS 2025, Aachen (IEEE Signal Processing Society).
PyTorch/TensorFlow Deep learning Differentiable programming
Part-time Lecturer Mar 2024 – Jun 2025
Université de Rennes · Rennes, France (On-site)
  • ACV - Video compression basics (ESIR 2nd year)
  • AI & Data Science - M2 EEEA and ISTIC
  • Python / C++ Programming - M2 SIVOS, ISTIC
Video Codec Research Intern Jan 2023 – Jul 2023
MediaKind · Cesson-Sévigné, France
  • Designed fast algorithms for VVC/HEVC intra/inter partitioning.
  • Built a prediction model to estimate the optimal split pattern for a given coding unit, reducing encoding complexity.
H.266 / VVC HEVC Partitioning C++
Video Research Intern May 2022 – Aug 2022
Haivision · Rennes, France
  • AV1 / VP9 transcoding evaluation for social network delivery pipelines.
  • Subjective and objective quality assessments (VMAF, PSNR, BD-Rate).
  • Real-time analysis on Intel platform with SVT optimizations.
  • Developed a full performance analysis tool (GStreamer / libx264) in Shell / Python / JS.
  • Integrated ML algorithms in C for content-aware preset prediction in H.264.
AV1 VP9 Python Shell Parallel Programming
Video Research Intern Jun 2021 – Aug 2021
Harmonic · France

Study of the impact of AI-based upscaling (Pixop) followed by downscaling on perceived quality and H.264/H.265 encoding performance.

Super-resolution H.264 / H.265 Perceptual Quality

🛠 Skills

Programming & Frameworks

Python (advanced)
C / C++ (advanced)
PyTorch/TensorFlow
VHDL

Video & Compression

H.264 / AVC (x264)
H.265 / HEVC
H.266 / VVC (VVenC)
AV1 / VP9
ABR Streaming
Bitrate Ladder

Machine Learning & AI

Deep Learning
CNN / U-Net
Differentiable Programming
Surrogate Gradients

📄 Publications

LoResNet - Light Codec-Aware Downscaling via Latent Distillation for Real-Time ABR Streaming December 2026
E. Pesnel, J. Le Tanou, M. Ropert, A. Roumy, T. Maugey

IEEE International Conference on Visual Communications and Image Processing (VCIP) 2026, Singapore

Conference paper Accepted
A Projection-Based Surrogate Gradient Interpretation for Neural Codec Wrappers June 2026
E. Pesnel, J. Le Tanou, M. Ropert, A. Roumy, T. Maugey

IEEE International Conference on Multimedia Signal Processing (MMSP) 2026

Conference paper Oral
SCALED - Surrogate Codec Approach for Learned End-to-end Downscaling December 2025
E. Pesnel, J. Le Tanou, M. Ropert, T. Maugey, A. Roumy

Picture Coding Symposium (PCS) 2025, Aachen - IEEE Signal Processing Society

Conference paper Poster

🏫 Teaching

AI Tools - PCA, SVM/SVR, Machine Learning Fall 2025
M2 SIVOS - ISTIC, Université de Rennes
Coding Tools - C++, OO Programming, Python Fall 2025
M2 SIVOS - ISTIC, Université de Rennes
ACV - Image & Entropy Coding Spring 2024 & 2025
ESIR, Université de Rennes - with A. Roumy (INRIA)

🌍 Languages

🇫🇷
French
Native
🇬🇧
English
TOEIC 935 / 990
🇪🇸
Spanish
Professional