talk-data.com

===== Inscription via EventBrite ======

Afin d’avoir une meilleure visibilité sur les personnes présentes ainsi que participer à une noble cause, nous demandons une participation symbolique qui sera entièrement reversée à l’institut Pasteur.

⚠️‼️ Toutes les inscriptions doivent se faire au travers de ce lien EventBrite. ===================================

🎉 Bonjour à vous!

Pour bien commencer l’année 2026 nous vous invitons à un meetup exceptionnel ✨!

❓ Du "ClickOps" sans trahir le GitOps, et une plateforme IA en production sur Kubernetes avec GPU : deux talks concrets pour ceux qui veulent aller au-delà de la théorie.

🎁 Pour ne rien gâcĥer il y aura quelques surprises: stickers, tee-shirts ... ainsi que 2 places à gagner pour le Cloud Native Days France 2026.

Nous tenons à remercier Scaleway pour l'accueil dans leurs bureaux à Paris!

📅 Découvrez l'agenda :

- 19.00 - Le GitOps ré-inventé : Corrigez la dérive d’état avant qu’elle ne se produise. (Scaleway)

- 19:40 - L’adoption massive de l’IA par les chercheurs de l’Institut Pasteur grâce à Kubernetes. (Pasteur / Enix)

- 20:15 - Food & Drinks 🍕☕️ Conversations 🗣️🙋

- 22:00 - Cloture du Meetup

===== Talks =====

Le GitOps ré-inventé : Corrigez la dérive d’état avant qu’elle ne se produise

Speaker:

• Damien Dassieu (Scaleway) Je suis un contributeur actif aux projets Kubernetes (Kubebuilder, controller-runtime, ...). Avec une expérience particulière dans le domaine du Platform Engineering acquise chez Orange et Scaleway dans les équipe Kubernetes as a Service, j’ai un rôle crucial en tant que développeur Go pour permettre la réalisation des idées de chacun. Je contribue actuellement au développement de Syngit, un outil qui comble la séparation entre les principes de GitOps et l’utilisation des outils impératifs pour la gestion de l'infrastructure.

Description:

Le GitOps est devenu une pratique fondamentale dans les environnements Kubernetes, offrant une approche déclarative et contrôlée par version de la gestion de l'infrastructure et des applications. Pourtant, les interactions concrètes avec les clusters reposent souvent sur des outils impératifs tels que kubectl, oc ou des interfaces web qui fournissent un retour utilisateur plus rapides et une expérience plus intuitive.

Ces outils impératifs valident et envoient des requêtes directement au serveur API de Kubernetes sans prendre en compte Git, ce qui casse le paradigme GitOps. Les conséquences sont le manque de visibilité, de reproductibilité et d'auditabilité.

Quelle est la pièce manquante pour allier ClickOps et GitOps ? Cette présentation introduit une nouvelle façon d’utiliser les outils impératifs en utilisant le GitOps. Ce mécanisme repose sur l’interception des requêtes vers l'API Kubernetes et en les transformant en commits Git. Nous explorerons comment Syngit s'intègre de manière transparente avec des outils GitOps tels que ArgoCD ou FluxCD pour préserver Git comme source de vérité même lorsque les utilisateurs interagissent avec les clusters de manière impérative.

---------

L’adoption massive de l’IA par les chercheurs de l’Institut Pasteur grâce à Kubernetes

Speakers:

• Stéphane Fournier (Institut Pasteur) Directeur des Systèmes d'Information à l'Institut Pasteur depuis 2018, Stéphane Fournier impulse la transformation numérique de la recherche biomédicale. Il a initié divers développements en IA en collaboration avec des équipes scientifiques expertes. Récemment, c'est la plateforme AI Tools qui concrétise cette approche. Basée sur une architecture hexagonale modulaire, elle permet le déploiement d'outils utilisant largement des LLMs publics, ou on-premise lorsque la souveraineté des données est en jeu.
• Jean-Baptiste Masson (Institut Pasteur) Physicien théoricien, Jean-Baptiste Masson est responsable de l’équipe Decision and Bayesian Computation (DBC-Épiméthée) à l’Institut Pasteur, une unité mixte Pasteur-Inria-CNRS-UPC. Ses recherches portent sur les principes physiques et computationnels qui organisent le traitement de l’information biologique, en lien avec l’environnement sensoriel. Elles couvrent notamment la chimiotaxie bactérienne, les stratégies de recherche olfactive chez les insectes et l’organisation des circuits neuronaux. Il mène également des travaux appliqués en algorithmique médicale (neuroAI), avec un fort accent sur des technologies déployables à l’hôpital, en particulier en imagerie médicale (DIVA et Genuage) et en analyse de données cliniques.
• Alexandre Buisine (Enix) Avec 15 ans d'expérience dans la technique et le management, il est un passionné d’open-source et de ce qui touche aux conteneurs. Il rejoint l'équipe Enix en 2018 comme directeur général, tout en restant impliqué dans les technos d’infrastructures réseaux et systèmes. Avant Enix, Alexandre a travaillé 3 ans au développement d’un service de réalité virtuelle. Il a également été CTO puis dirigeant de SmartJog, dans les domaines de l'audio-visuel, la transmission satellite, le développement logiciel, les réseaux IP, la distribution de contenus sur Internet.

Description:

L'Institut Pasteur a déployé un écosystème d**'IA générative** pour ses 2 500 chercheurs, orchestré sur Kubernetes avec trois environnements (dev, prod, GPU) et une chaîne CI/CD GitLab.

L’architecture repose sur deux proxies LiteLLM (cloud et on-premise) unifiant l’accès aux modèles (Mistral, Anthropic, OpenAI côté cloud ; Perplexity, Qwen, Llama côté on-premise), et assurant le suivi des coûts par token et le load balancing.

Cette couche d'abstraction alimente trois catégories de services : LibreChat pour le dialogue généraliste enrichi de serveurs MCP, AI Tools pour les usages métiers spécialisés (analyse de publications scientifiques), et des assistants de codage 100% on-premise (Qwen-Coder, Roo-Coder) intégrés aux IDE.

Kubernetes nous permet de remplacer un modèle open source en quelques minutes et d'ajuster dynamiquement le scaling à la charge.

Les défis techniques incluent la gestion de GPU hétérogènes dans un cluster Talos sécurisé, la compatibilité entre anciens matériels et systèmes DGX B200 NVIDIA, et les migration Kubernetes à chaud sans downtime.

Entre besoins métier et décisions d'architecture, nous retracerons le parcours complet : du prototype développé par un chercheur jusqu'au déploiement en production, en expliquant le rôle des échanges entre scientifiques et équipe infrastructure dans cette plateforme cloud-native.

Enfin, nous partagerons les coulisses opérationnelles : architecture, gestion du GPU Operator, stratégies de scaling, et surtout les leçons tirées de nos 6 mois d'exploitation en production.

===== Consentement à la photographie/vidéo =====

Nous pourrons prendre des photos et vidéos lors de l'événement et les utiliser sur les réseaux sociaux et dans des supports promotionnels. Participer au meetup implique que vous consentez à ce que nous prenions des photos et vidéos de vous.

🎦 Ce meetup sera retransmis en live lsur nos canaux traditionnels (YouTube & LinkedIn)

📰 Au programme: Chaque projet Fabric bute sur la même question : comment découper mes workspaces ? Ce choix n’est jamais neutre : il conditionne la gestion des capacities, l’organisation par Domaine (métiers… ou pas 😉) et, in fine, la gouvernance de votre plateforme. Dans cette session, nous vous proposons une démarche pour réfléchir à cette organisation, en croisant contraintes techniques et choix organisationnels. Nous aborderons aussi la question clé de la délégation d’administration, qu’elle s’opère via les Domaines ou les Capacités. Nous ferons également un pas de côté avec l’approche Data Mesh, pour voir comment la notion de Data Product peut inspirer la structuration de vos environnements Fabric. Une session pragmatique pour repartir avec une grille de lecture claire et éviter le chaos des workspaces improvisés.

Le meetup sera présenté par 🤵‍♂️Jean-Pierre Riehl

ℹ️Nous rappelons que le Club Fabric est une association et que toute l'équipe d'organisation est bénévole. La communauté, c'est vous, c'est nous. Le fondement d'un club utilisateurs c'est le partage de connaissances et d'expérience. Si vous avez des idées, des remarques, des envies, de la motivation, contactez-nous et contribuez 😉

Bonjour à tous,

Nous organisons pour la première fois une journée de "Super meet up" avec des anciens speakers qui reviennent pour nous parler de leur dernières avancées. Cette journée aura lieu le 9 décembre au CNIT. Elle se déroule au sein de la conférence apidays où ils nous offrent une salle.

Bonne nouvelle : nous avons 80 places à offrir à la communauté via ce lien.

Pour assister à la conférence, il faut absolument prendre une place sur le site apidays. Le pass vous donne également droit au reste de la conférence les autres jours.

Pour les talks nous sommes très contents d'annoncer :

• AGO : Maxime Thoonsen Founder
• Partoo : Thibault Renouf CO-CEO
• Docker : David Gageot Senior Principal Software Engineer et Jean-Laurent de Morlhon VP Software Engineering
• Linkup : Boris Toledano COO
• 360Learning : Marie Douriez Lead Machine Learning Engineer et Adrien Gros AI Product Manager
• Brut : Alexandre Lavallee Ingénieur IA et Iris Izraylevich Responsable pôle data
• Mirakl : Fanny Guélin Senior Product Manager et Pierre Lourdelet Data Scientist
• Wavestone: Ari Kouts Manager Innovation & AI
• Bangcast : Nikolay Martynov
• Theodo : Thomas Walter CTO France
• LeBonCoin : Yoann C. Engineering Manager et Yann Lemonnier Staff Machine Learning Engineer

Vous avez également des places à 30% de réduction ici s'il n'en reste plus de gratuites via ce lien.

Hâte de vous y retrouver,

La team Generative AI France

Présentation sur les solutions de monitoring pour Microsoft Fabric: pourquoi le monitoring, les niveaux de supervision (Tenant, Capacity, Gateway, Item), outils et bonnes pratiques (FUAM, FCA), cas d’usage et retours d’expérience, et conseils pour améliorer la gouvernance et la performance de votre environnement de données.

Hear talks from experts on the latest topics in AI, ML, and computer vision on November 13.

Date and Location

Nov 13, 2025 9 AM Pacific Online. Register for the Zoom!

Copy, Paste, Customize! The Template Approach to AI Engineering

Most AI implementations fail because teams treat prompt engineering as ad-hoc experimentation rather than systematic software engineering, leading to unreliable systems that don't scale beyond proof-of-concepts. This talk demonstrates engineering practices that enable reliable AI deployment through standardized prompt templates, systematic validation frameworks, and production observability.

Drawing from experience developing fillable prompt templates currently being validated in production environments processing thousands of submissions, I'll share how Infrastructure as Code principles apply to LLM workflows, why evaluation metrics like BLEU scores are critical for production reliability, and how systematic failure analysis prevents costly deployment issues. Attendees will walk away with understanding of practical frameworks for improving AI system reliability and specific strategies for building more consistent, scalable AI implementations.

About the Speaker

Jeanne McClure is a postdoctoral scholar at NC State's Data Science and AI Academy with expertise in systematic AI implementation and validation. Her research transforms experimental AI tools into reliable production systems through standardized prompt templates, rigorous testing frameworks, and systematic failure analysis. She holds a PhD in Learning, Design and Technology with additional graduate work in data science.

Multimodality with Biases: Understand and Evaluate VLMs for Autonomous Driving with FiftyOne

Do your VLMs really see danger? With FiftyOne, I’ll show you how to understand and evaluate vision-language models for autonomous driving — making risk and bias visible in seconds. We’ll compare models on the same scenes, reveal failures and edge cases, and you’ll see a simple dashboard to decide which data to curate and what to adjust. You’ll leave with a clear, practical, and replicable method to raise the bar for safety.

About the Speaker

Paula Ramos has a PhD in Computer Vision and Machine Learning, with more than 20 years of experience in the technological field. She has been developing novel integrated engineering technologies, mainly in Computer Vision, robotics, and Machine Learning applied to agriculture, since the early 2000s in Colombia.

The Heart of Innovation: Women, AI, and the Future of Healthcare

This session explores how Artificial Intelligence is transforming healthcare by enhancing diagnosis, treatment, and patient outcomes. It highlights the importance of diverse and female perspectives in shaping AI solutions that are ethical, empathetic, and human-centered. We will discuss key applications, current challenges, and the future potential of AI in medicine. It’s a forward-looking conversation about how innovation can build a healthier world.

About the Speaker

Karen Sanchez is a Postdoctoral Researcher at the Center of Excellence for Generative AI at King Abdullah University of Science and Technology (KAUST), Saudi Arabia. Her research focuses on AI for Science, spanning computer vision, video understanding, and privacy-preserving machine learning. She is also an active advocate for diversity and outreach in AI, contributing to global initiatives that connect researchers and amplify underrepresented voices in technology.

Language Diffusion Models

Autoregressive models (ARMs) are widely regarded as the cornerstone of large language models (LLMs). Challenge this notion by introducing LLaDA, a diffusion model trained from scratch under the pre-training and supervised fine-tuning (SFT) paradigm. LLaDA models distributions through a forward data masking process and a reverse process, parameterized by a vanilla Transformer to predict masked tokens.

Optimizing a likelihood bound provides a principled generative approach for probabilistic inference. Across extensive benchmarks, LLaDA demonstrates strong scalability, outperforming self-constructed ARM baselines. Remarkably, LLaDA 8B is competitive with strong LLMs like LLaMA3 8B in in-context learning and, after SFT, exhibits impressive instruction-following abilities in case studies such as multi-turn dialogue.

About the Speaker

Jayita Bhattacharyya is an AI/ML Nerd with a blend of technical speaking & hackathon wizardry! Applying tech to solve real-world problems. The work focus these days is on generative AI. Helping software teams incorporate AI into transforming software engineering.

Hear talks from experts on the latest topics in AI, ML, and computer vision on November 13.

Date and Location

Nov 13, 2025 9 AM Pacific Online. Register for the Zoom!

Copy, Paste, Customize! The Template Approach to AI Engineering

Most AI implementations fail because teams treat prompt engineering as ad-hoc experimentation rather than systematic software engineering, leading to unreliable systems that don't scale beyond proof-of-concepts. This talk demonstrates engineering practices that enable reliable AI deployment through standardized prompt templates, systematic validation frameworks, and production observability.

Drawing from experience developing fillable prompt templates currently being validated in production environments processing thousands of submissions, I'll share how Infrastructure as Code principles apply to LLM workflows, why evaluation metrics like BLEU scores are critical for production reliability, and how systematic failure analysis prevents costly deployment issues. Attendees will walk away with understanding of practical frameworks for improving AI system reliability and specific strategies for building more consistent, scalable AI implementations.

About the Speaker

Jeanne McClure is a postdoctoral scholar at NC State's Data Science and AI Academy with expertise in systematic AI implementation and validation. Her research transforms experimental AI tools into reliable production systems through standardized prompt templates, rigorous testing frameworks, and systematic failure analysis. She holds a PhD in Learning, Design and Technology with additional graduate work in data science.

Multimodality with Biases: Understand and Evaluate VLMs for Autonomous Driving with FiftyOne

Do your VLMs really see danger? With FiftyOne, I’ll show you how to understand and evaluate vision-language models for autonomous driving — making risk and bias visible in seconds. We’ll compare models on the same scenes, reveal failures and edge cases, and you’ll see a simple dashboard to decide which data to curate and what to adjust. You’ll leave with a clear, practical, and replicable method to raise the bar for safety.

About the Speaker

Paula Ramos has a PhD in Computer Vision and Machine Learning, with more than 20 years of experience in the technological field. She has been developing novel integrated engineering technologies, mainly in Computer Vision, robotics, and Machine Learning applied to agriculture, since the early 2000s in Colombia.

The Heart of Innovation: Women, AI, and the Future of Healthcare

This session explores how Artificial Intelligence is transforming healthcare by enhancing diagnosis, treatment, and patient outcomes. It highlights the importance of diverse and female perspectives in shaping AI solutions that are ethical, empathetic, and human-centered. We will discuss key applications, current challenges, and the future potential of AI in medicine. It’s a forward-looking conversation about how innovation can build a healthier world.

About the Speaker

Karen Sanchez is a Postdoctoral Researcher at the Center of Excellence for Generative AI at King Abdullah University of Science and Technology (KAUST), Saudi Arabia. Her research focuses on AI for Science, spanning computer vision, video understanding, and privacy-preserving machine learning. She is also an active advocate for diversity and outreach in AI, contributing to global initiatives that connect researchers and amplify underrepresented voices in technology.

Language Diffusion Models

Autoregressive models (ARMs) are widely regarded as the cornerstone of large language models (LLMs). Challenge this notion by introducing LLaDA, a diffusion model trained from scratch under the pre-training and supervised fine-tuning (SFT) paradigm. LLaDA models distributions through a forward data masking process and a reverse process, parameterized by a vanilla Transformer to predict masked tokens.

Optimizing a likelihood bound provides a principled generative approach for probabilistic inference. Across extensive benchmarks, LLaDA demonstrates strong scalability, outperforming self-constructed ARM baselines. Remarkably, LLaDA 8B is competitive with strong LLMs like LLaMA3 8B in in-context learning and, after SFT, exhibits impressive instruction-following abilities in case studies such as multi-turn dialogue.

About the Speaker

Jayita Bhattacharyya is an AI/ML Nerd with a blend of technical speaking & hackathon wizardry! Applying tech to solve real-world problems. The work focus these days is on generative AI. Helping software teams incorporate AI into transforming software engineering.

Hear talks from experts on the latest topics in AI, ML, and computer vision on November 13.

Date and Location

Nov 13, 2025 9 AM Pacific Online. Register for the Zoom!

Copy, Paste, Customize! The Template Approach to AI Engineering

Most AI implementations fail because teams treat prompt engineering as ad-hoc experimentation rather than systematic software engineering, leading to unreliable systems that don't scale beyond proof-of-concepts. This talk demonstrates engineering practices that enable reliable AI deployment through standardized prompt templates, systematic validation frameworks, and production observability.

Drawing from experience developing fillable prompt templates currently being validated in production environments processing thousands of submissions, I'll share how Infrastructure as Code principles apply to LLM workflows, why evaluation metrics like BLEU scores are critical for production reliability, and how systematic failure analysis prevents costly deployment issues. Attendees will walk away with understanding of practical frameworks for improving AI system reliability and specific strategies for building more consistent, scalable AI implementations.

About the Speaker

Jeanne McClure is a postdoctoral scholar at NC State's Data Science and AI Academy with expertise in systematic AI implementation and validation. Her research transforms experimental AI tools into reliable production systems through standardized prompt templates, rigorous testing frameworks, and systematic failure analysis. She holds a PhD in Learning, Design and Technology with additional graduate work in data science.

Multimodality with Biases: Understand and Evaluate VLMs for Autonomous Driving with FiftyOne

Do your VLMs really see danger? With FiftyOne, I’ll show you how to understand and evaluate vision-language models for autonomous driving — making risk and bias visible in seconds. We’ll compare models on the same scenes, reveal failures and edge cases, and you’ll see a simple dashboard to decide which data to curate and what to adjust. You’ll leave with a clear, practical, and replicable method to raise the bar for safety.

About the Speaker

Paula Ramos has a PhD in Computer Vision and Machine Learning, with more than 20 years of experience in the technological field. She has been developing novel integrated engineering technologies, mainly in Computer Vision, robotics, and Machine Learning applied to agriculture, since the early 2000s in Colombia.

The Heart of Innovation: Women, AI, and the Future of Healthcare

This session explores how Artificial Intelligence is transforming healthcare by enhancing diagnosis, treatment, and patient outcomes. It highlights the importance of diverse and female perspectives in shaping AI solutions that are ethical, empathetic, and human-centered. We will discuss key applications, current challenges, and the future potential of AI in medicine. It’s a forward-looking conversation about how innovation can build a healthier world.

About the Speaker

Karen Sanchez is a Postdoctoral Researcher at the Center of Excellence for Generative AI at King Abdullah University of Science and Technology (KAUST), Saudi Arabia. Her research focuses on AI for Science, spanning computer vision, video understanding, and privacy-preserving machine learning. She is also an active advocate for diversity and outreach in AI, contributing to global initiatives that connect researchers and amplify underrepresented voices in technology.

Language Diffusion Models

Autoregressive models (ARMs) are widely regarded as the cornerstone of large language models (LLMs). Challenge this notion by introducing LLaDA, a diffusion model trained from scratch under the pre-training and supervised fine-tuning (SFT) paradigm. LLaDA models distributions through a forward data masking process and a reverse process, parameterized by a vanilla Transformer to predict masked tokens.

Optimizing a likelihood bound provides a principled generative approach for probabilistic inference. Across extensive benchmarks, LLaDA demonstrates strong scalability, outperforming self-constructed ARM baselines. Remarkably, LLaDA 8B is competitive with strong LLMs like LLaMA3 8B in in-context learning and, after SFT, exhibits impressive instruction-following abilities in case studies such as multi-turn dialogue.

About the Speaker

Jayita Bhattacharyya is an AI/ML Nerd with a blend of technical speaking & hackathon wizardry! Applying tech to solve real-world problems. The work focus these days is on generative AI. Helping software teams incorporate AI into transforming software engineering.

Hear talks from experts on the latest topics in AI, ML, and computer vision on November 13.

Date and Location

Nov 13, 2025 9 AM Pacific Online. Register for the Zoom!

Copy, Paste, Customize! The Template Approach to AI Engineering

Most AI implementations fail because teams treat prompt engineering as ad-hoc experimentation rather than systematic software engineering, leading to unreliable systems that don't scale beyond proof-of-concepts. This talk demonstrates engineering practices that enable reliable AI deployment through standardized prompt templates, systematic validation frameworks, and production observability.

Drawing from experience developing fillable prompt templates currently being validated in production environments processing thousands of submissions, I'll share how Infrastructure as Code principles apply to LLM workflows, why evaluation metrics like BLEU scores are critical for production reliability, and how systematic failure analysis prevents costly deployment issues. Attendees will walk away with understanding of practical frameworks for improving AI system reliability and specific strategies for building more consistent, scalable AI implementations.

About the Speaker

Jeanne McClure is a postdoctoral scholar at NC State's Data Science and AI Academy with expertise in systematic AI implementation and validation. Her research transforms experimental AI tools into reliable production systems through standardized prompt templates, rigorous testing frameworks, and systematic failure analysis. She holds a PhD in Learning, Design and Technology with additional graduate work in data science.

Multimodality with Biases: Understand and Evaluate VLMs for Autonomous Driving with FiftyOne

Do your VLMs really see danger? With FiftyOne, I’ll show you how to understand and evaluate vision-language models for autonomous driving — making risk and bias visible in seconds. We’ll compare models on the same scenes, reveal failures and edge cases, and you’ll see a simple dashboard to decide which data to curate and what to adjust. You’ll leave with a clear, practical, and replicable method to raise the bar for safety.

About the Speaker

Paula Ramos has a PhD in Computer Vision and Machine Learning, with more than 20 years of experience in the technological field. She has been developing novel integrated engineering technologies, mainly in Computer Vision, robotics, and Machine Learning applied to agriculture, since the early 2000s in Colombia.

The Heart of Innovation: Women, AI, and the Future of Healthcare

This session explores how Artificial Intelligence is transforming healthcare by enhancing diagnosis, treatment, and patient outcomes. It highlights the importance of diverse and female perspectives in shaping AI solutions that are ethical, empathetic, and human-centered. We will discuss key applications, current challenges, and the future potential of AI in medicine. It’s a forward-looking conversation about how innovation can build a healthier world.

About the Speaker

Karen Sanchez is a Postdoctoral Researcher at the Center of Excellence for Generative AI at King Abdullah University of Science and Technology (KAUST), Saudi Arabia. Her research focuses on AI for Science, spanning computer vision, video understanding, and privacy-preserving machine learning. She is also an active advocate for diversity and outreach in AI, contributing to global initiatives that connect researchers and amplify underrepresented voices in technology.

Language Diffusion Models

Autoregressive models (ARMs) are widely regarded as the cornerstone of large language models (LLMs). Challenge this notion by introducing LLaDA, a diffusion model trained from scratch under the pre-training and supervised fine-tuning (SFT) paradigm. LLaDA models distributions through a forward data masking process and a reverse process, parameterized by a vanilla Transformer to predict masked tokens.

Optimizing a likelihood bound provides a principled generative approach for probabilistic inference. Across extensive benchmarks, LLaDA demonstrates strong scalability, outperforming self-constructed ARM baselines. Remarkably, LLaDA 8B is competitive with strong LLMs like LLaMA3 8B in in-context learning and, after SFT, exhibits impressive instruction-following abilities in case studies such as multi-turn dialogue.

About the Speaker

Jayita Bhattacharyya is an AI/ML Nerd with a blend of technical speaking & hackathon wizardry! Applying tech to solve real-world problems. The work focus these days is on generative AI. Helping software teams incorporate AI into transforming software engineering.

Hear talks from experts on the latest topics in AI, ML, and computer vision on November 13.

Date and Location

Nov 13, 2025 9 AM Pacific Online. Register for the Zoom!

Copy, Paste, Customize! The Template Approach to AI Engineering

Most AI implementations fail because teams treat prompt engineering as ad-hoc experimentation rather than systematic software engineering, leading to unreliable systems that don't scale beyond proof-of-concepts. This talk demonstrates engineering practices that enable reliable AI deployment through standardized prompt templates, systematic validation frameworks, and production observability.

Drawing from experience developing fillable prompt templates currently being validated in production environments processing thousands of submissions, I'll share how Infrastructure as Code principles apply to LLM workflows, why evaluation metrics like BLEU scores are critical for production reliability, and how systematic failure analysis prevents costly deployment issues. Attendees will walk away with understanding of practical frameworks for improving AI system reliability and specific strategies for building more consistent, scalable AI implementations.

About the Speaker

Jeanne McClure is a postdoctoral scholar at NC State's Data Science and AI Academy with expertise in systematic AI implementation and validation. Her research transforms experimental AI tools into reliable production systems through standardized prompt templates, rigorous testing frameworks, and systematic failure analysis. She holds a PhD in Learning, Design and Technology with additional graduate work in data science.

Multimodality with Biases: Understand and Evaluate VLMs for Autonomous Driving with FiftyOne

Do your VLMs really see danger? With FiftyOne, I’ll show you how to understand and evaluate vision-language models for autonomous driving — making risk and bias visible in seconds. We’ll compare models on the same scenes, reveal failures and edge cases, and you’ll see a simple dashboard to decide which data to curate and what to adjust. You’ll leave with a clear, practical, and replicable method to raise the bar for safety.

About the Speaker

Paula Ramos has a PhD in Computer Vision and Machine Learning, with more than 20 years of experience in the technological field. She has been developing novel integrated engineering technologies, mainly in Computer Vision, robotics, and Machine Learning applied to agriculture, since the early 2000s in Colombia.

The Heart of Innovation: Women, AI, and the Future of Healthcare

This session explores how Artificial Intelligence is transforming healthcare by enhancing diagnosis, treatment, and patient outcomes. It highlights the importance of diverse and female perspectives in shaping AI solutions that are ethical, empathetic, and human-centered. We will discuss key applications, current challenges, and the future potential of AI in medicine. It’s a forward-looking conversation about how innovation can build a healthier world.

About the Speaker

Karen Sanchez is a Postdoctoral Researcher at the Center of Excellence for Generative AI at King Abdullah University of Science and Technology (KAUST), Saudi Arabia. Her research focuses on AI for Science, spanning computer vision, video understanding, and privacy-preserving machine learning. She is also an active advocate for diversity and outreach in AI, contributing to global initiatives that connect researchers and amplify underrepresented voices in technology.

Language Diffusion Models

Autoregressive models (ARMs) are widely regarded as the cornerstone of large language models (LLMs). Challenge this notion by introducing LLaDA, a diffusion model trained from scratch under the pre-training and supervised fine-tuning (SFT) paradigm. LLaDA models distributions through a forward data masking process and a reverse process, parameterized by a vanilla Transformer to predict masked tokens.

Optimizing a likelihood bound provides a principled generative approach for probabilistic inference. Across extensive benchmarks, LLaDA demonstrates strong scalability, outperforming self-constructed ARM baselines. Remarkably, LLaDA 8B is competitive with strong LLMs like LLaMA3 8B in in-context learning and, after SFT, exhibits impressive instruction-following abilities in case studies such as multi-turn dialogue.

About the Speaker

Jayita Bhattacharyya is an AI/ML Nerd with a blend of technical speaking & hackathon wizardry! Applying tech to solve real-world problems. The work focus these days is on generative AI. Helping software teams incorporate AI into transforming software engineering.

Hear talks from experts on the latest topics in AI, ML, and computer vision on November 13.

Date and Location

Nov 13, 2025 9 AM Pacific Online. Register for the Zoom!

Copy, Paste, Customize! The Template Approach to AI Engineering

Most AI implementations fail because teams treat prompt engineering as ad-hoc experimentation rather than systematic software engineering, leading to unreliable systems that don't scale beyond proof-of-concepts. This talk demonstrates engineering practices that enable reliable AI deployment through standardized prompt templates, systematic validation frameworks, and production observability.

Drawing from experience developing fillable prompt templates currently being validated in production environments processing thousands of submissions, I'll share how Infrastructure as Code principles apply to LLM workflows, why evaluation metrics like BLEU scores are critical for production reliability, and how systematic failure analysis prevents costly deployment issues. Attendees will walk away with understanding of practical frameworks for improving AI system reliability and specific strategies for building more consistent, scalable AI implementations.

About the Speaker

Jeanne McClure is a postdoctoral scholar at NC State's Data Science and AI Academy with expertise in systematic AI implementation and validation. Her research transforms experimental AI tools into reliable production systems through standardized prompt templates, rigorous testing frameworks, and systematic failure analysis. She holds a PhD in Learning, Design and Technology with additional graduate work in data science.

Multimodality with Biases: Understand and Evaluate VLMs for Autonomous Driving with FiftyOne

Do your VLMs really see danger? With FiftyOne, I’ll show you how to understand and evaluate vision-language models for autonomous driving — making risk and bias visible in seconds. We’ll compare models on the same scenes, reveal failures and edge cases, and you’ll see a simple dashboard to decide which data to curate and what to adjust. You’ll leave with a clear, practical, and replicable method to raise the bar for safety.

About the Speaker

Paula Ramos has a PhD in Computer Vision and Machine Learning, with more than 20 years of experience in the technological field. She has been developing novel integrated engineering technologies, mainly in Computer Vision, robotics, and Machine Learning applied to agriculture, since the early 2000s in Colombia.

The Heart of Innovation: Women, AI, and the Future of Healthcare

This session explores how Artificial Intelligence is transforming healthcare by enhancing diagnosis, treatment, and patient outcomes. It highlights the importance of diverse and female perspectives in shaping AI solutions that are ethical, empathetic, and human-centered. We will discuss key applications, current challenges, and the future potential of AI in medicine. It’s a forward-looking conversation about how innovation can build a healthier world.

About the Speaker

Karen Sanchez is a Postdoctoral Researcher at the Center of Excellence for Generative AI at King Abdullah University of Science and Technology (KAUST), Saudi Arabia. Her research focuses on AI for Science, spanning computer vision, video understanding, and privacy-preserving machine learning. She is also an active advocate for diversity and outreach in AI, contributing to global initiatives that connect researchers and amplify underrepresented voices in technology.

Language Diffusion Models

Autoregressive models (ARMs) are widely regarded as the cornerstone of large language models (LLMs). Challenge this notion by introducing LLaDA, a diffusion model trained from scratch under the pre-training and supervised fine-tuning (SFT) paradigm. LLaDA models distributions through a forward data masking process and a reverse process, parameterized by a vanilla Transformer to predict masked tokens.

Optimizing a likelihood bound provides a principled generative approach for probabilistic inference. Across extensive benchmarks, LLaDA demonstrates strong scalability, outperforming self-constructed ARM baselines. Remarkably, LLaDA 8B is competitive with strong LLMs like LLaMA3 8B in in-context learning and, after SFT, exhibits impressive instruction-following abilities in case studies such as multi-turn dialogue.

About the Speaker

Jayita Bhattacharyya is an AI/ML Nerd with a blend of technical speaking & hackathon wizardry! Applying tech to solve real-world problems. The work focus these days is on generative AI. Helping software teams incorporate AI into transforming software engineering.

Hear talks from experts on the latest topics in AI, ML, and computer vision on November 13.

Date and Location

Nov 13, 2025 9 AM Pacific Online. Register for the Zoom!

Copy, Paste, Customize! The Template Approach to AI Engineering

Most AI implementations fail because teams treat prompt engineering as ad-hoc experimentation rather than systematic software engineering, leading to unreliable systems that don't scale beyond proof-of-concepts. This talk demonstrates engineering practices that enable reliable AI deployment through standardized prompt templates, systematic validation frameworks, and production observability.

Drawing from experience developing fillable prompt templates currently being validated in production environments processing thousands of submissions, I'll share how Infrastructure as Code principles apply to LLM workflows, why evaluation metrics like BLEU scores are critical for production reliability, and how systematic failure analysis prevents costly deployment issues. Attendees will walk away with understanding of practical frameworks for improving AI system reliability and specific strategies for building more consistent, scalable AI implementations.

About the Speaker

Jeanne McClure is a postdoctoral scholar at NC State's Data Science and AI Academy with expertise in systematic AI implementation and validation. Her research transforms experimental AI tools into reliable production systems through standardized prompt templates, rigorous testing frameworks, and systematic failure analysis. She holds a PhD in Learning, Design and Technology with additional graduate work in data science.

Multimodality with Biases: Understand and Evaluate VLMs for Autonomous Driving with FiftyOne

Do your VLMs really see danger? With FiftyOne, I’ll show you how to understand and evaluate vision-language models for autonomous driving — making risk and bias visible in seconds. We’ll compare models on the same scenes, reveal failures and edge cases, and you’ll see a simple dashboard to decide which data to curate and what to adjust. You’ll leave with a clear, practical, and replicable method to raise the bar for safety.

About the Speaker

Paula Ramos has a PhD in Computer Vision and Machine Learning, with more than 20 years of experience in the technological field. She has been developing novel integrated engineering technologies, mainly in Computer Vision, robotics, and Machine Learning applied to agriculture, since the early 2000s in Colombia.

The Heart of Innovation: Women, AI, and the Future of Healthcare

This session explores how Artificial Intelligence is transforming healthcare by enhancing diagnosis, treatment, and patient outcomes. It highlights the importance of diverse and female perspectives in shaping AI solutions that are ethical, empathetic, and human-centered. We will discuss key applications, current challenges, and the future potential of AI in medicine. It’s a forward-looking conversation about how innovation can build a healthier world.

About the Speaker

Karen Sanchez is a Postdoctoral Researcher at the Center of Excellence for Generative AI at King Abdullah University of Science and Technology (KAUST), Saudi Arabia. Her research focuses on AI for Science, spanning computer vision, video understanding, and privacy-preserving machine learning. She is also an active advocate for diversity and outreach in AI, contributing to global initiatives that connect researchers and amplify underrepresented voices in technology.

Language Diffusion Models

Autoregressive models (ARMs) are widely regarded as the cornerstone of large language models (LLMs). Challenge this notion by introducing LLaDA, a diffusion model trained from scratch under the pre-training and supervised fine-tuning (SFT) paradigm. LLaDA models distributions through a forward data masking process and a reverse process, parameterized by a vanilla Transformer to predict masked tokens.

Optimizing a likelihood bound provides a principled generative approach for probabilistic inference. Across extensive benchmarks, LLaDA demonstrates strong scalability, outperforming self-constructed ARM baselines. Remarkably, LLaDA 8B is competitive with strong LLMs like LLaMA3 8B in in-context learning and, after SFT, exhibits impressive instruction-following abilities in case studies such as multi-turn dialogue.

About the Speaker

Jayita Bhattacharyya is an AI/ML Nerd with a blend of technical speaking & hackathon wizardry! Applying tech to solve real-world problems. The work focus these days is on generative AI. Helping software teams incorporate AI into transforming software engineering.

Welcome to DSF Sandbox Sessions!

Top tech talks and masterclasses from the best in the industry. Every month, and completely free. The catch? There isn’t one. Just a monthly dose of epic content to power up your data passion!

Event details 📆 Date: Thursday 6th November 2025 ⏰ Time: 2 – 4 PM GMT/9–11 AM EST 💡 Topic: Data Vibe Coding with AI: Prompting Strategies That Actually Work 🗣️ Speakers: Greg Michaelson and Jean-Dominique Mercury from Zerve.

How do you get the best results when coding with AI? Should you craft one big, complex prompt and hope for the best, or start small and refine as you go? In this hands-on session, we’ll explore “vibe coding,” the art of shaping your AI prompts so that your code agent works with you instead of against you. You’ll learn practical techniques for structuring prompts, steering outputs, and avoiding common dead ends. Along the way, we’ll build something real together in Zerve, giving you the chance to try it out live, ask questions, and see how different prompting styles play out in practice. While we’ll be working inside Zerve for the live demo, the prompting principles and vibe coding techniques you’ll learn apply to any AI coding tool or workflow.

Watch live and you’ll also earn a Data Vibe Coding 101 Certificate of Completion from The Data Science Festival and Zerve.

Join us for an interactive two-hour code-along hosted with the Data Science Festival on November 6th 2025. Whether you’re a data scientist, analyst, or engineer, you’ll come away with a set of prompting strategies you can immediately apply in your own workflow. These skills are platform-agnostic and designed to improve how you collaborate with AI in any environment.

We’ll cover:

• When to go broad vs. when to go step-by-step with your prompts
• How to use corrective prompts to guide your AI agent toward the right solution
• Common pitfalls that derail AI-assisted coding—and how to avoid them

This isn’t a passive lecture. You’ll sign up, bring a dataset of your choice, and follow along as we build something from scratch. You’ll get hands-on practice inside Zerve, learn techniques you can use anywhere, and leave with both a finished project and a deeper sense of how to get into the “flow” of coding with AI.

This event is for anyone curious about vibe coding for data, from complete beginners to seasoned tech enthusiasts. No prior coding experience needed, just curiosity. In the second half, step into the sandbox: bring a dataset you want to explore, and get hands-on with support from two Zerve experts. They’ll help you play, learn, and build, answering questions and levelling up your skills at any stage.

Save time by creating a free account on Zerve that you’ll use for data vibe coding during the workshop. Simply go to https://www.zerve.ai/?utm_source=dsf-vibes from a desktop/laptop, click ‘Try Zerve Now’ and complete the sign up process with a Google account or your email address and a password ahead of the event.

Registering for the event: Click the 'Sign up here!' button on the specific event page following the link below. Once you have completed the registration form, you will be emailed a link to join the webinar. You will also receive a reminder link one week, one day and one hour prior to the event.

PLEASE NOTE: Clicking 'attend' on Meetup does not register you for the event. You will need to register for the event on the link provided below to receive a joining link. If you do not, you will not be able to join the event.

Click here to sign up for this specific event

Please note the time zone when you book this event.

Animation et débat sur les nouveautés Power BI et l’actualité.

De la tradition culinaire à l’innovation digitale, Saint Jean démontre comment la donnée peut devenir un véritable levier de performance.

Présentation du sujet : Java Meets Spark - Retour d’expérience pour comprendre, utiliser et créer de la valeur avec Apache Spark.