22 sessions
Reading groupFederated Learning: Machine Learning on the EdgePushed backThe speaker challenged the assumption that conventional cloud-based machine learning pipelines are sufficient, arguing that federated learning is needed for personalization, privacy, security, and cost reasons.16:44
PodcastData Mesh: Data Quality Control Mechanism for MLOps?ClaimA data product should include its transformations, consumption experience, service-level agreements, documentation, and information about how trustworthy and current it is.8:54
PodcastPractitioners Guide to MLOpsClaimData and model management supports governance, auditability, traceability, compliance, shareability, reusability, and discoverability of machine learning artifacts.13:43
PodcastA Journey in Scaling MLClaimA golden path can improve time to value, reproducibility, security, and compliance, but there should be different golden paths for different machine-learning applications.19:29
PodcastThe Shipyard: Lessons Learned While Building an ML PlatformClaimThe platform team treats model engineers as users and relies on trust, feedback, and iterative improvements rather than treating compliance errors as malicious behavior.15:17
PodcastModel Monitoring in Practice: Top TrendsClaimExplainability and model monitoring are becoming as important to AI systems as security became after organizations experienced data breaches.8:50
MeetupFLOps with Scaleout's Open-core PlatformPushed backFederated learning is not automatically the answer to privacy challenges when data cannot be standardized or even a small sample cannot be shared centrally.37:51
PodcastSpeed Up Data-Driven ValueClaimDelina Ivanova says previous experience in finance, consulting, operations, strategy, data governance and data ethics helped her become more effective at leading a data team.5:25
PodcastJust Fetch the Data and then...ClaimLexisNexis Risk Solutions must restrict the data it makes available according to legal permissions, which can produce different answers for different customers.12:02
PodcastScaling Machine Learning with Data MeshClaimData mesh began mainly as an approach for analytics data, where each data product should have a valid use case, add value, and be trustworthy and discoverable.5:37
PodcastData Engineering for MLClaimA data contract gives data a clear owner, a defined purpose, trustworthy delivery, and a form that consumers understand.19:55
PodcastTrustworthy Machine LearningPushed backKrishnaram Kenthapadi questioned whether machine learning systems are different from complex machines and medicines that people trust without understanding how they work.9:07
PodcastReliable Machine LearningClaimNiall Murphy says organizations should take privacy, access control, and subgroup disadvantage seriously when they begin using machine learning.50:28
PodcastManaging Machine Learning ProjectsClaimSimon Thompson says that scalability requirements should be handled when they are real, while manageability, governability, usability, transparency, and explainability are required for production systems.28:22
MeetupUnleashing Sensitive Datasets with Distributed Data ScienceClaimDifferential privacy adds noise to data or results to provide privacy protection.14:20
MeetupCreative AI: Using ML to Create Art, Music, and JokesClaimSuyash Joshi says AI-generated art raises questions about attribution, copyright, data permissions, and the value of art.15:27
MeetupDriving ML Data Quality with Data ContractsClaimData is critical to model performance because effective models require sound, trusted, and measurable data.6:40
PodcastBuilding Threat Detection Systems: An MLE's PerspectivePushed backJeremy Jordan clarified that rules and machine learning usually work together in cybersecurity rather than rules simply being replaced by machine learning.8:41
PodcastExplainability in the MLOps CycleClaimDattaraj Rao says his three main research areas at Persistent are knowledge platforms and knowledge graphs, responsible AI and privacy-preserving AI, with MLOps closely related to responsible AI.13:28
