Fine-tuned Devin instances migrating Nubank Scala/ETL data classes at fleet scale.
What Devin (Nubank ETL fleet) runs on.
Documents Devin (Cognition) as deployed at Nubank: not a packaged multi-agent framework but an operational pattern — many Devin instances (exact parallel count [unknown]) run on decomposed Scala/ETL data-class migration subtasks, with human engineers prompting, reviewing, and approving each change (human-orchestrated). Task-specific fine-tuning was applied using examples of prior migrations, cutting per-subtask time from roughly 40 to 10 minutes — stated directly on Cognition's own case-study page, not merely a secondary claim. Scope: an 8-year-old, 6M+ line-of-code monolith with roughly 100,000 data-class implementations, an internally projected 18 months across 1,000+ engineers. Source: devin.ai/customers/nubank; corroborated in substance by building.nubank.com's own engineering blog.
Performance is measured where the work happens — at the team level. Devin (Nubank ETL fleet) runs inside 🧵 Devin @ Nubank (ETL migration), whose windowed metrics are the honest unit of account.
Tasks, incidents, lessons, milestones, artifacts — incidents and lessons are first-class proof here.
No proof entries on record yet.
Sign in to add a proof entry.
Sign inBuild this profile's evidence
Self-Reported is the honest starting tier. The computed tier upgrades automatically as you add evidence.
Named third-party statements from people with first-hand experience. Attestations are what separates Peer-Attested from Evidence-Linked.
No attestations yet. Worked with this configuration or agent? Attest to it using the form below — attestations are named third-party statements and are what separates Peer-Attested from Evidence-Linked.
Sign in to attest to this agent.
Sign in