Sonia Tabti

Cali Intelligences develops and deploys Computer Vision solutions for the retail industry. I ponctually help them designing these solutions and building their Computer Vision models faster.

I can supervise data projects from presales and framing to execution and deployment. I can advise on data collection, hardware selection, labeling and modeling strategies. I enjoy challenging use case formulations, solving problems and mentoring teammates. With my academic background and my pedagogic skills, I can increase data science and AI awareness within your non-technical and more technical teams. I can also define a Research and Development AI strategy to help you tackle future challenges efficiently.

Dans le cadre du DE Expert IA (formation continue délivrant un diplôme équivalent à un Master 2 pour des professionnels déjà dans la vie active), j'ai pu intervenir et donner des cours sur les thèmes suivants: - Cycle de vie d'un projet d'IA en entreprise - Défis de chaque étape de ce cycle de vie - L'importance d'une bonne stratégie de gouvernance de données - Régulations : RGPD, IA act - Sensibilisation aux risques d'un usage peu raisonné des assistants d'IA - Mises en pratique: extraction de données de documents PDFs et développement d'un système de RAG

- It’s a company aiming to build batteries for electric cars - Use cases vary from predictive maintenance on tabular data to chatbots and Computer Vision - I supervised a team of 3 data-scientists and 1 ML-Ops - I mentored a junior data scientist - I improved the agile methodology of the team - I defined a long term roadmap for a top priority use case dealing with defect detection on images - I contributed to the popularization of data labeling tools to the business team - I did the technical framing of 2 future critical use cases (one about image quality control and one for an advanced RAG system)

AI Forensics is an NGO holding Big Tech companies and their platforms and algorithms accountable. To do so, algorithmic auditing pipelines are developed by their team. The data collected during those audits requires labeling. My main mission was to design with my teammates the labeling workflow and develop the labeling interfaces. I also contributed to improving their project management rituals and research methodology.

I had the opportunity to review the TEADAL project.

-Laboratory: GREYC, Team: Image processing. -Fundings: ANR-FEDER GRAPHSIP and CPER NUMNIE -Topic: graph signal processing (images and colored 3D point clouds) with variationnal approaches (variants of Total Variation on graphs) -Applications: Heritage preservation. -Collaborators: Julien Rabin and Abderrahim Elmoataz. -Organized a mini-symposium at the SIAM'18 conference in Bologna, Italy. -Publications + seminars.

-At the DIETI. -Area: Synthetic Aperture Radar (SAR) image processing (satellite images). -Proposed a new sparse-coding method adapted to SAR images with a classification map based on coefficient of variation. -Application: Despeckling. -Collaborators: Luisa Verdoliva, Giovanni Poggi

Pendant la rédaction de ma thèse à Télécom ParisTech, j'ai occupé un poste d'ATER à temps plein d'un an à Paris Descartes où j'intervienais en tant que chargée de TD dans différentes matières de la licence de Mathématiques Informatique et Applications. Au premier semestre, j'étais chargée d'un groupe de TD de Maths et Calcul 1 (L1), et d'un groupe d'Algèbre linéaire 3 (L2). Au second semestre, j'ai poursuivi les TD de Maths et Calculs 1 (L1) et encadré deux étudiants en projet (L2).

Distribution de flyers.

• Gestion des stocks de fournitures et de publicité • Classement de dossiers • Saisie de clients dans la base de données de gestion des formations • Relance de factures impayées • Recherche de nouveaux stagiaires, formation des nouveaux arrivants