Devin for Databricks migrations
Databricks migrations in
weeks instead of months
Spend less time migrating and more time building on Databricks. Devin is the autonomous AI software engineer from Cognition. Point Devin at your legacy warehouse, ETL, or notebooks and it converts the code to Databricks, runs every converted file against your workspace, and reconciles the data before you cut over.
- Schema
- Row counts
- Aggregates
- Row-by-row diff
Databricks Data + AI World Tour
Meet with Cognition at the Databricks Data + AI World Tour
Meet with us on the road in Sydney, Munich, London, Toronto, New York City, Milan, and Paris.
View World Tour events- SydneySep 16
- MunichNov 4
- LondonNov 11
- TorontoNov 17
- New York CityNov 19
- MilanNov 24
- ParisDec 8
Why migrations stall
They take too long
Manual, line-by-line conversion of SQL, stored procedures, and pipelines is slow work, and the cost runs into the millions.
They consume your best engineers
The people who know the legacy system get pulled off the roadmap for a year of repetitive translation.
They carry cutover risk
Tangled dependencies, logic errors, and data drift surface at go-live, when they are most expensive to fix.
How Devin runs the migration
Your engineers review the work instead of doing it line by line, so the migration finishes in days and the team gets back to building on the Databricks Data + AI Platform.
- 01
Agentic transpilation
Devin converts Redshift, Snowflake, Teradata, Oracle, SQL Server, and Informatica code (SQL, stored procedures, ETL) to Databricks SQL, PySpark, and Apache Spark Declarative Pipelines (SDP). It then executes every converted file on your Databricks workspace and fixes what fails.
- 02
Agentic reconciliation
After migration, Devin compares every source table with its Databricks copy: schema, row counts, aggregates, then a row-by-row diff. Every missing row and changed cell is found and documented, whether the table has ten thousand rows or a hundred million.
- 03
Parallelization at scale
Hundreds of migration tasks run at the same time, each in its own isolated Devin session. Devin coordinates the whole workload, so months of manual work fit into days.
Fortune 500 energy company | Migration to Databricks Unity Catalog
Devin migrated and validated a 142-object platform in 8 days
What moved
ETL notebooks, a claims change-data-capture pipeline, and scheduled jobs that feed Power BI, actuarial, and financial reporting.
What Devin did
Devin created the schemas, built the pipelines, and resolved issues on its own as they came up.
Where it landed
The migration finished well ahead of the traditional manual estimate, and the team is now planning the next phase.
More migration results
Customer storiesWhat Devin migrates
ETL pipelines
Informatica, Talend, SSIS
Databricks Workflows and SDPHadoop MapReduce and Hive
Optimized PySpark
SQL and data warehouses
Oracle PL/SQL, T-SQL, Teradata BTEQ
Databricks SQLRedshift, Snowflake, BigQuery
Databricks SQL and Unity Catalog
Legacy code
PySpark 2.x
PySpark 3.x on Databricks RuntimeJupyter and Zeppelin notebooks
Databricks Notebooks
Inside your Databricks
governance boundary
Devin operates within your Databricks workspace governance boundary, inheriting Unity Catalog policies and lineage tracking natively.
Connect Devin to DatabricksOne auditable identity
Devin acts as a service principal you created, so every API call, query, and job run shows up in Databricks audit logs and Unity Catalog lineage under that identity, not under an engineer's personal token.
Unity Catalog decides what Devin can touch
OAuth decides whether Devin can authenticate. Unity Catalog grants and workspace permissions decide what it can read or change. You can start read-only in production, give Devin a sandbox catalog to build in, and widen scope only after you have seen how it behaves.
A path to no stored secret
With OIDC token federation, Devin never stores a Databricks token or client secret. Each session exchanges a 60-second Devin identity token for a short-lived Databricks OAuth token.
Why Devin
Model agnostic by design
Devin is a compound AI system that accesses every frontier model in one platform, preventing vendor lock-in. Tasks are routed to the best-fit model.
Optimized for value generation
You pay for measured value. Devin's industry-first $10M AI Productivity Guarantee ties spend directly to measured engineering value. Sessions are categorized by type of work, converted to human-equivalent hours, and verified at year-end. If value falls short, Cognition funds the gap up to $10M.
Intelligence optimized for cost. Devin continuously improves the cost for outcomes. Devin Fusion, which routes each task across multiple models, delivers ~35% better cost-efficiency while maintaining frontier performance.
About the guaranteeForward-deployed transformation partner
Dedicated Forward Deployed Engineers co-own business outcomes with customers. Devin teams provide white-glove onboarding, onsite enablement, and structured measurement of ROI.
Cloud agent maturity
Move from augmenting individuals to multiplying teams. Delegate work to fleets of parallel agents in secure, isolated Linux and Windows environments that complete the full verification loop, with knowledge that compounds across teams.
Enterprise security and governance
6 of the 8 US Global Systemically Important Banks (G-SIBs) use Devin. Data is encrypted in transit and at rest, with customer-managed keys and private networking available. Inputs and outputs are your IP, attribution filtering is available, and there is no training on enterprise customer data by default. Governance includes fine-grained access and spend controls, role-based permissions through custom roles and orgs, comprehensive audit logs, and multiple deployment options.
Draw against your Databricks Universal Commit
Devin is transactable on the Databricks Marketplace. If you have a Databricks Universal Commit, you can draw against it when you purchase Devin.
Devin by Cognition on the Databricks Marketplace