Feature stores have become a critical piece of ML infrastructure, but many teams still struggle with the 'why' and 'how'. In this episode, Lucas and Luna dig into a specific case: how a mid-size e-commerce company cut model development time by 40 percent after adopting Feast, an open-source feature store. They discuss the pain of feature engineering duplication, the shift from notebooks to production pipelines, and the surprising lesson that feature stores are as much about data governance as they are about speed. If you're building ML systems or wondering whether a feature store is worth the investment, this episode gives you a concrete decision framework. No hype, just what actually worked.

#FeatureStore #Feast #MLInfrastructure #MachineLearning #DataEngineering #DataGovernance #FeatureEngineering #ModelDevelopment #MLOps #OpenSource #Ecommerce #DataPlatform #BusinessAndTechnology #DataTeams #FexingoBusiness #BusinessPodcast #DataBusiness #Episode112

Keep every episode free: buymeacoffee.com/fexingo

In this episode of The Data Business Podcast, Lucas and Luna explore how data teams are applying workforce intelligence — combining HR data, productivity metrics, and engagement signals — to predict employee retention. They focus on the case of a mid-sized tech company that reduced attrition by 22% using a churn model built on Slack activity patterns, manager feedback frequency, and project diversity. Lucas explains the difference between vanity metrics like average tenure and leading indicators like 'time since last promotion conversation.' Luna questions whether this crosses into surveillance territory, and they discuss the ethical guardrails teams need to put in place. The conversation also covers how data engineers are building pipelines from Workday, Jira, and GitHub to feed these models, and why the biggest challenge isn't the algorithm — it's getting clean, consent-gated data. Tune in for a practical look at an emerging analytics use case at the intersection of people ops and data engineering.

#WorkforceIntelligence #PeopleAnalytics #EmployeeRetention #HRData #PredictiveModeling #DataEngineering #SlackAnalytics #EthicalAI #PeopleOps #DataPipelines #Workday #Jira #GitHub #ChurnModel #Business #Technology #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

Episode 110 of The Data Business Podcast. Lucas and Luna explore how data teams are adopting the OpenTable file format to break free from proprietary table formats like Delta Lake and Apache Iceberg. They discuss the real-world case of a mid-sized e-commerce company that switched to OpenTable for cross-platform analytics, the technical and organizational challenges involved, and the trade-offs between performance and portability. The conversation covers how OpenTable simplifies data sharing across Spark, Flink, and Trino, why it reduces vendor lock-in, and what it means for data platform teams. Along the way, they touch on the tactical decisions around catalog integration, compaction, and schema evolution. A practical look at why interoperability is becoming the new default for data lakehouse architectures in mid-2026.

#OpenTable #DataFormat #DataInteroperability #Lakehouse #ApacheSpark #ApacheFlink #Trino #DataArchitecture #DataPlatform #VendorLockIn #SchemaEvolution #DataCatalog #BusinessAndTechnology #DataTeam #DataEngineering #DataLake #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

Episode 109 of The Data Business Podcast dives into the rise of data observability — the practice of monitoring data pipelines for freshness, volume, schema, and quality. Lucas and Luna explore how Monte Carlo and Bigeye are leading the category, and why data teams are moving from reactive alerting to proactive observability. They break down a key metric: mean time to detection (MTTD) dropping from days to minutes. The episode also covers the shift from in-house monitoring scripts to full-stack observability platforms, and how companies like WeWork and Fox adopted these tools post-2020. If you're building or running data infrastructure, this episode explains why observability is becoming as essential as monitoring was for DevOps.

#DataObservability #MonteCarloData #Bigeye #DataQuality #DataEngineering #DataPipelines #MTTD #DataFreshness #VolumeMonitoring #SchemaChange #DataLineage #SlackAlerting #RealTimeMonitoring #WeWork #Fox #BarrMoses #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

Episode 108 of The Data Business Podcast explores why data teams are converging on lakehouse architecture for stream processing. Lucas and Luna unpack the shift from Lambda architectures to unified stream-batch on Apache Iceberg, using a real-world case: a mid-sized logistics company that cut pipeline complexity by 60 percent after moving to a lakehouse with Apache Flink. They discuss the trade-offs around latency, cost, and tooling maturity, and why the lakehouse model is winning for teams that need both real-time analytics and historical accuracy. Specific numbers: sub-second latency on 10 million events per day, storage costs down 35 percent, and query performance within 80 percent of a dedicated streaming database. No marketing fluff, just what's happening inside data engineering teams in mid-2026.

#Lakehouse #StreamProcessing #ApacheFlink #ApacheIceberg #DataArchitecture #RealTimeAnalytics #LambdaArchitecture #DataEngineering #BusinessAndTechnology #FexingoBusiness #BusinessPodcast #DataInfrastructure #PipelineOptimization #StreamBatchUnified #DataPlatform #EventDriven #DataStreaming #CostOptimization

Keep every episode free: buymeacoffee.com/fexingo

Episode 107 of The Data Business Podcast explores why data teams are increasingly turning to reverse ETL to push analytics directly into customer-facing products. Lucas and Luna examine a case study of a mid-market SaaS company that used reverse ETL to embed usage metrics into its dashboard, reducing churn by 18% over two quarters. They break down the technical stack, the cultural shift required from data teams to think like product engineers, and the trade-offs between real-time syncs and batch updates. The episode also touches on how reverse ETL changes the data team's role from a cost center to a revenue driver. If you're building data infrastructure that touches customers directly, this one's for you.

#ReverseETL #CustomerFacingAnalytics #DataEngineering #DataInfrastructure #SaaS #ProductLedGrowth #EmbeddedAnalytics #ChurnReduction #DataTeamCulture #RealTimeData #DataPipelines #BusinessIntelligence #CustomerSuccess #DataMonetization #OperationalAnalytics #Business #Technology #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

Lucas and Luna dive into the shift from batch to real-time feature engineering in ML inference pipelines. They discuss a specific case: a mid-size e-commerce company that cut fraud-detection latency from 15 minutes to under 200 milliseconds by using streaming feature stores with Apache Flink. The episode covers the architectural change—moving from daily batch jobs to event-time feature computation—and the trade-offs in consistency, cost, and monitoring. Lucas shares a data point: the team reduced fraud losses by 12% in the first quarter after the switch. They also talk about why most teams still batch features, and when real-time makes sense. A concrete look at the operational reality of serving features at inference time.

#RealTimeFeatures #FeatureEngineering #MLInference #StreamingData #ApacheFlink #FraudDetection #FeatureStore #OnlineInference #DataEngineering #MachineLearning #Latency #EventTime #DataInfrastructure #Business #Technology #FexingoBusiness #BusinessPodcast #TheDataBusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

In this episode, Lucas and Luna dive into data clean rooms—a technology that lets companies collaborate on customer data without exposing raw PII. They break down how retailers like Target partner with payment networks like Mastercard to measure ad effectiveness and uncover marketing attribution without ever sharing individual customer records. The hosts walk through the technical architecture: how differential privacy noise is added, how Snowflake and AWS are competing on native clean room features, and why the post-cookie era has accelerated adoption. They also explore a real-world case: a regional grocery chain using a clean room to identify overlapping loyalty customers with a CPG brand while keeping purchase histories private. Lucas argues clean rooms are becoming the default trust layer for data sharing in regulated industries, while Luna questions whether current solutions are truly auditable or just black boxes. The episode closes with a look at what's next—clean rooms as a service and the tension between privacy and data utility.

#DataCleanRooms #PrivacyCompliance #Snowflake #AWS #Mastercard #Target #RetailAnalytics #MarketingAttribution #DifferentialPrivacy #CustomerData #AdTech #CPG #LoyaltyPrograms #DataSharing #RegTech #BusinessAndTechnology #FexingoBusiness #BusinessPodcast

Keep every episode free: buymeacoffee.com/fexingo

In this episode of The Data Business Podcast, Lucas and Luna dive into how data teams are leveraging dbt's Semantic Layer to power embedded analytics inside customer-facing products. They break down a specific case: a mid-market SaaS company that replaced a homegrown API layer with dbt's Semantic Layer to serve consistent metrics across their web app and customer dashboards. Lucas walks through the technical architecture — how the Semantic Layer sits between the data warehouse and the application, using MetricFlow under the hood to define dimensions and measures in YAML. Luna pushes on the operational trade-offs: what happens when a customer needs a custom metric that the semantic definition doesn't cover? They discuss the governance benefits — column-level lineage baked into every metric — and the real cost savings from eliminating redundant ETL. The conversation lands on where this pattern works best and where it still struggles. No fluff, just a clear-headed look at one team's decision to standardize their analytics layer.

#DataBusiness #BusinessAndTechnology #FexingoBusiness #BusinessPodcast #dbt #SemanticLayer #EmbeddedAnalytics #MetricStore #DataEngineering #DataInfrastructure #AnalyticsEngineering #DataGovernance #DataLineage #DataModeling #SaaS #ProductAnalytics #DataArchitecture #Fexingo

Keep every episode free: buymeacoffee.com/fexingo

Episode 103 of The Data Business Podcast explores data compaction for Apache Iceberg tables—a technique that merges small files into larger ones to improve query performance and reduce storage costs. Lucas and Luna break down why compaction matters, using a real example from a mid-size e-commerce company that cut its Athena query costs by 40 percent after adopting a nightly compaction routine. They discuss trade-offs like write amplification and the role of tools like Apache Spark and Iceberg's built-in optimize procedure. The episode also touches on the rise of tiered storage and how compaction fits into modern data lakehouse architectures. Perfect for data engineers and analytics leaders managing large Iceberg deployments.

#DataCompaction #ApacheIceberg #DataLakehouse #QueryPerformance #StorageOptimization #DataEngineering #BigData #ApacheSpark #AmazonAthena #CostOptimization #DataInfrastructure #OpenTableFormats #Analytics #BusinessTechnology #DataTeam #FexingoBusiness #BusinessPodcast #DataOps

Keep every episode free: buymeacoffee.com/fexingo