13 sessions
MeetupReal-time Feature Pipelines, A Personal HistoryPushed backHendrik Brackmann says the usefulness of storing dynamic model outputs as features depends on reuse, training design, and system boundaries rather than having one universal answer.45:02
PodcastMachine Learning Feature Store Panel DiscussionPushed backWhether a company needs to buy or build a feature store depends on its use cases and maturity rather than having a single universal answer.9:29
MeetupMachine Learning Design Patterns for MLOpsClaimA feature store is useful when multiple models need to reuse the same calculated features, but many use cases only need a library.13:04
Meetup2 tools = 90% operational MLClaimFeature stores provide abstractions over different data architectures and simplify decisions involved in building MLOps pipelines.15:29
PodcastMachine Learning at AtlassianClaimAtlassian initially had scattered data analysts who often acted as full-stack data scientists, but it did not yet have machine-learning-powered product features.13:20
MeetupFrom Idea to Production MLClaimLex Beattie says explainable AI can be used to debug models and check whether features or neurons are behaving as expected.37:19
MeetupBuilding ML Blocks with Kubeflow Orchestration with Feature StorePushed backAniruddha Choudhury says Feast is not a workflow scheduler, data warehouse, pipeline orchestrator, feature engineering tool, or model-serving product.16:10
MeetupWhat MLOps Has Taught MeClaimEwan Nicolson demonstrates a locally run recommender system with streaming data ingestion, a data lake, collaborative filtering, embeddings, an approximate-nearest-neighbors index, a feature store, business rules, and an API.8:55
PodcastThe Future of ML and Data PlatformsClaimMichael Del Balso says the safest way to build a platform is to start with basic capabilities, help teams use them, and turn repeated bespoke work into platform features.21:12
MeetupDoing MLOpsClaimA feature store can hold cleaned, transformed, scaled, and documented data for both machine learning and business intelligence.15:11
MeetupFeast Feature Store Deep DiveClaimFeast is an open-source feature store.4:51
PodcastLinkedIn Job RecommendationsClaimThe team refactored its models so that personalization was handled in activity features rather than inside each model's learned weights.5:49
PodcastML Stepping Stones: Challenges & Opportunities for CompaniesClaimJohn Crousse says not every feature containing machine learning has a business case for constant monitoring or incremental improvement.15:47