I’m a Principal Media Streaming Engineer at Disney. I design the Content-Aware ABR Stacks behind Disney+ and work on Objective & Perceptual Video Quality, Dialogue Clarity and Automated QC. Before that I architected DIRECTV’s 4K/HDR/Atmos launch, co-developed DTS:X and ITU-T G.719, the first full-band audio standard for video conferencing.
2003
Engineering Degree and MSc
🇫🇷 Lannion
2003–2007
PhD in DSP & Computer Science
🇫🇷 Grenoble / Lannion
2007–2011
Research Engineer
🇸🇪 Stockholm
2011–2015
Senior R&D Engineer
🇺🇸 Calabasas
2015–2021
Principal Systems Architect
🇺🇸 El Segundo
2021–Present
Principal Media Streaming Engineer
🇺🇸 Glendale
For 23 years I’ve worked on how media looks and sounds: codec standards at Ericsson (ITU-T G.719), DTS:X at DTS, the DIRECTV 4K/HDR/Atmos launch at AT&T, and now content-aware ABR stacks, perceptual quality, dialogue clarity and automated QC for Disney+.
Teams bring me three questions: is the quality good enough, why is this content or system behaving this way, and what should we do about it. More and more, my answer is automation and machine learning that carry a trained ear and eye across an entire catalog.
15%less Disney+ VOD delivery, on average
I design the content-aware (CAE) ABR stacks behind Disney+, one for each format: SDR, 4K HDR, IMAX and now 9:16 vertical video in UHD and FHD, SDR and HDR. Per-variant objective and perceptual scoring cut VOD delivery by 15% on average, weighted by viewing, and up to 40% on some of the most-watched titles. Content-specific presets save hundreds of thousands a year in CDN costs.
9:16vertical video: sound that follows the crop, in development
Disney+ vertical video is mostly cropped from 16:9 masters, but cinematic mixes are made for the full widescreen frame. I invented a way to use the cropping data to re-balance the soundfield, so the sound follows what stays in the 9:16 picture. It is still in development, not yet in production. Patent pending.
100%of Disney+ devices, no new licensing
A patented, backward-compatible way to make dialogue easier to understand on every Disney+ device. Production integration is complete on Skywalker Sound’s CODA Edge (on-premise, containerized). With Fraunhofer, I co-developed DIS, an objective score for dialogue intelligibility. Paper at the AES Show 2026 in Nashville, Tennessee.
24×faster than real time, 30 languages
I’m building automated pipelines that score every rendition and send only the segments that need a human to a reviewer. The dub and subtitle language checks, built on open-source ML models, flagged no file falsely on a 187-file validation set and run about 24× faster than real time.
50%+less egress and fewer production issues
At AT&T I pushed leadership to adopt CMAF across live and cloud DVR: one set of segments instead of duplicate MPEG-TS and fMP4 copies. Egress costs and production issues fell by more than half.
At DTS, DIRECTV and now Disney, I train QC teams to run subjective reviews. Today I’m teaching the ingest and QC teams to use and improve the automated QC pipeline.
Both interns I supervised during my industrial (CIFRE) PhD at Orange Labs went on to PhDs. At AT&T, an intern I mentored was hired full time, and I led a Cal State LA student team on a KPI-emulator project.
Founding member of the CMAF Industry Forum, with contributions to ITU-T G.719, the ATSC 3.0 audio submission that became DTS:X, DASH-IF guidelines, Apple’s HLS authoring spec and CTA WAVE. Member of the Alliance for Open Media (AV1, AV2, IAMF) and SVTA. Reviewer for AES and IEEE.
Industry and Standards Work