Marie Couvé

Skills: Gestion de projet · Volunteer Recruiting

Latitudes est une association qui sensibilise aux enjeux de la Tech. J'anime plusieurs ateliers : - Bataille de l'IA : sensibilisation aux enjeux de l'IA générative lors d'un atelier de 2h pour les étudiant.es, les citoyen.nes et les professionnel.les - Bataille de la Tech : sensibilisation aux enjeux de la tech lors d'un atelier de 2h pour les étudiant.es, les citoyen.nes et les professionnel.les - Future of Tech : sensibilisation aux enjeux de la tech et aux métiers du numérique lors d'un atelier de 2h pour les collégien.nes et lycéen.nes N'hésitez pas à me contacter pour un atelier dans votre entreprise ou votre école !

Axionable is a sustainable and responsible Artificial Intelligence consulting firm My achievements: ★ Deliver sustainable and responsible AI projects for our customers from scoping to production ★ Be a referent on responsible AI at Axionable: continuous improvement of our AI delivery methodology, renewal of labels and certificates, monitoring, training of Axionable teams, animation of the tribe responsible AI, and construction of new offers ★ Disseminate responsible AI best practices to our clients during audits and trusted AI training ★ Internal diversity and inclusion awareness ________ My missions in detail: 1️⃣ Audit and recommendations on the responsible AI strategy of a large media group: - Analyze the development methodologies of an AI project and documentation from the responsible AI perspective - Make recommendations on responsible AI methodology and tools to anticipate European regulation (AI Act) 2️⃣ Responsible AI training for the AI team of a large nuclear group: - Imagine a training path covering all responsible AI pillars and combining theory and practice - Build training materials and notebooks - Lead 3 training days for 10 people 3️⃣ Audit, recommendations, and implementation of a methodology for a large construction group: - Analyze the development and documentation practices of an AI project and its compliance with the LNE standard - Implement a responsible AI methodology accompanied by documentation for compliance with responsible AI best practices and existing repositories 4️⃣ CSR awareness for a software publisher: - Build training materials and Klaxoon activities - Facilitate training sessions

My achievements: ★ Develop an algorithm for automatic analysis of financial and non-financial reports of companies ★ Improve internal methodologies and infuse responsible AI best practices ★ Obtain the AI certificate from the LNE and the "responsible and trusted AI" label from Labelia to ensure the good integration of the best practices of responsible AI ★ Internal diversity and inclusion awareness ________ My missions in detail: 1️⃣ Automatic analysis of financial and non-financial reports of companies: - Label data as a learning base - Analyze the state of the art for automatic answers to multiple choice questions (Question Answering, Classification, NER...) - Implement, evaluate, and optimize models - Track training traceability with ML Flow Tracking 2️⃣ Obtaining the IA certificate by the LNE (National Metrology and Testing Laboratory): - Bring our AI delivery methodology in line with the requirements of the LNE standard - Create documentation templates to meet LNE repository requirements - Perform internal white audits before the final audit - Conduct an audit and answer questions from LNE auditors - Ensure communication with auditors until certification 3️⃣ Obtaining the Labelia label "responsible and trusted AI": - Bring our AI delivery methodology and responsible AI tools into compliance - Lead the labeling process from start to finish - Ensure communication with Labelia until the label is obtained - Participate in the communication around the label 4️⃣ Dashboard for monitoring financial and non-financial KPIs: - Define needs with different stakeholders - Collect data and calculate defined KPIs - Propose and implement selected graphs in DataStudio

End of studies internship as a Data Scientist Predictive maintenance for a major player in the nuclear industry : - Analysis, cleaning and spelling correction of textual data (Symspell) - Clustering in different kind of failures - Prediction of the next failure Enrichment of the internal tool for trustworthy AI: - Benchmark of AI’s carbon footprint calculation tools - Test the different tools and compare computation methods and results - Propose a methodology to be integrated into our internal tool - Make mock-ups of the carbon impact monitoring visuals of an AI project from its development to its launch in production