Data migrations have long been the costly, painful bottleneck of modernization, dragging on for years and carrying multi-million-dollar price tags. But AI is already flipping that script.

Gleb Mezhanskiy, founder & CEO of Datafold, who has been deep in the trenches of AI-driven migrations, joins us on the Data T podcast to unpack how AI automation is finally making it realistic to move off entrenched legacy ETL tools. Drawing on his years as a hands-on data engineer and PM (including leading a massive migration initiative at Lyft), Gleb explains where migrations go off the rails: millions of lines of legacy code and lack of automation lead to massive human time sunk into reconciliation and QA. He shares how Datafold attacks the problem end-to-end, using agents and data diffing to translate and validate at scale, so teams can “lift and shift” quickly, then refactor with confidence on modern stacks like Snowflake and Coalesce.

Beyond migrations, Gleb and podcast host, Coalesce co-founder and CEO Armon Petrossian, dig into how AI is reshaping data engineering itself. They argue AI won’t replace great engineers; it elevates them, shifting work from tedious rewrites to higher-leverage design, governance, and outcomes. The teams that master AI-assisted workflows, evaluation, and modern patterns will widen the gap, moving from weeks of manual effort to rapid, continuous delivery.

Key Topics

Data migration challenges
The role of AI in data migrations
Limitations of generic AI tools and LLM models for migration projects
How AI is reshaping data engineering roles
Predictions for data infrastructure
AI for productivity

Gleb Mezhanskiy on Rewriting the Migration Playbook with AI