Gleb Mezhanskiy

Evri, the UK's largest parcel delivery company (~1B parcels/year, 28K+ couriers), faced a daunting modernization: 1,200+ legacy objects across SAP HANA, Talend, and Power BI — all powering near-real-time operations with sub-10-minute SLAs. A traditional outsourced rewrite was projected to take 2–3 years. Hard contractual deadlines on both legacy platforms made that a non-starter. Datafold is a Databricks partner delivering outcome-based migrations at a fixed price with a guaranteed timeline, thanks to end-to-end AI automation. In this session, Evri's Chief Data Officer, Harvinder Atwal, and Datafold CEO, Gleb Mezhanskiy, share how Evri migrated and modernized all data processing on Databricks, leveraging Datafold's AI Migration Agent and Knowledge Graph for value-level validation across thousands of datasets. Both legacy platforms were decommissioned on time, in under a year, with zero disruption to operations.

With expensive contracts up for renewal, Evri faced the challenge of migrating 1,000 SAP HANA assets and 200+ Talend jobs to Databricks. This talk will cover how we transformed SAP HANA and Talend workflows into modern Databricks pipelines through AI-powered translation and validation -- without months of manual coding. We'll cover:- Techniques for handling SAP HANA's proprietary formats- Approaches for refactoring incremental pipelines while ensuring dashboard stability- The technology enabling automated translation of complex business logic- Validation strategies that guarantee migration accuracye'll share real examples of SAP HANA stored procedures transformed into Databricks code and demonstrate how we maintained 100% uptime of critical dashboards during the transition. Join us to discover how AI is revolutionizing what's possible in enterprise migrations from GUI-based legacy systems to modern, code-first data platforms.

In this session, Gleb Mezhanskiy, CEO of Datafold, will share innovative strategies for automating the conversion of legacy transformation code (i.e., stored procedures) to dbt models, a crucial step in modernizing your data infrastructure.