Skip to content
all episodes
Research Insights Made Simple · episode 28

The Data Platform in 2026: From DWH to Lakehouse and AI Agents

August 24, 20261:37:08
Written summary
Share
TelegramLinkedIn

Episode participants

  • Alexander Polomodov

    host

  • Nikolay Golov

    guest · Product Director · Tengri Data

    Nikolay Golov is Product Director at Tengri Data, formerly led data platforms at Avito and ManyChat, and teaches at Harbour.Space.

  • Alexander Filatov

    guest · Head of Engineering · Tengri Data

    Alexander Filatov leads engineering at Tengri Data. He previously spent almost ten years developing Avito's DWH and worked on its 2022–2025 migration from Vertica to Trino, Iceberg, and S3.

What we discussed

With Nikolay Golov and Alexander Filatov, the episode returns to data platforms: what Tengri Data built over the past year, how the platform works, and what pilots and sales revealed.

The discussion moves from the OLTP–OLAP boundary and the limits of classic MPP warehouses to an S3-and-Iceberg lakehouse, lightweight isolated compute, and the multiyear cost of legacy migration.

The conclusion is that AI agents multiply both productivity and risky workload, so a data platform needs strict permissions, reliable isolation, and proof that a query was executed and its result actually read.

Data platformsPlatform engineeringDatabasesAI in SDLC