At RMC BFM ADS (Altice Media), I developed a modular Python ETL architecture to replace legacy Talend infrastructure. The new routines cut processing times by 80% and eliminated Talend license costs.
I also developed serverless workflows for video ad transcoding and audio-to-video production, using AWS services to process, validate, and distribute media. For audience intelligence, I built an AWS Bedrock semantic matching pipeline for TV schedules and replay content, reaching a 95.1% matching rate with Claude 3 Haiku.
Earlier, at BPCE-SI, I developed and maintained Big Data pipelines on Hadoop and contributed to a Cloudera-to-GCP migration. At IMPERIUM and 3W Media, I worked on production AI and machine-learning projects including speech transcription, ad-spot recognition, sentiment analysis, and computer vision.

