Iceberg, ClickHouse®, and AI with Shivji Kumar Jha | Unevenly Distributed Ep. 8

🚀 Welcome to the eighth episode of Unevenly Distributed, a podcast exploring the future of real-time data lakes and the tradeoffs shaping modern analytics. In this episode, Shivji kumar Jha joins Altinity's Josh Lee (Open Source Developer Advocate) to discuss Iceberg, ClickHouse®, and AI. ________________________________________________________________ 🎧 About Unevenly Distributed Data is being reimagined in real time. In each episode of Unevenly Distributed, we dive deep with builders and practitioners to explore the technologies, tradeoffs, and frameworks redefining real-time data lakes. 📌 Subscribe here for future episodes → https://altinity.com/unevenly-distrib... ________________________________________________________________ 🔗 About Altinity: Website → https://altinity.com/ LinkedIn →   / altinity   GitHub → https://github.com/Altinity ________________________________________________________________ 1:12 — What has changed recently in the data ecosystem that made ClickHouse + Iceberg compelling? 2:02 — What does a unified data platform mean in practice? Is it marketing hype? 4:30 — What changed in your thinking over the last five years? 5:25 — Where does OLAP / ClickHouse shine, and where does it struggle? 6:42 — What mistakes do people make when getting started with ClickHouse? 7:46 — How do you help people think differently about transactions and OLAP? 8:37 — How was the experience combining ClickHouse and Postgres to support Iceberg? 11:02 — Why Iceberg? What’s exciting there? 13:42 — Can Iceberg bring Postgres and ClickHouse together around shared data? 14:44 — What does the data layer need to provide to support AI? 20:42 — What breaks if people think about AI workloads in old-fashioned ways? 23:23 — Is ClickHouse + Iceberg + Postgres enough, or do we need more? 25:15 — What trade-offs are people missing when they chase open tools? 29:27 — Does AI give open source more advantage by helping people analyze unfamiliar code? 32:21 — Have AI tools lowered the barrier for junior engineers to contribute to open source? 32:35 — Does that create problems for open-source maintainers? 34:54 — What will the data stack look like three years from now? 41:20 — What should more engineers be experimenting with today? 43:49 — How do we make AI-agent apps as reliable as traditional software? 44:24 — What gaps exist around AI infrastructure, observability, and agent systems? #datalake #datalakehouse #OpenLakehouse #dataanalytics #techpodcast #podcast #realtimedata #opensource #dataplatform #AI #apacheiceberg #clickhouse