Sergey Sergyenko's wroclove.rb 2022 talk. Defines data management as a holistic discipline broader than database management, ETL, data analytics or data science, spanning structure, architecture, shaping, protection and transformation of data. Surveys the variety of data-related professions (data engineer, data tester, data architect, data analyst, etc.) and cites a report listing Ruby among the top-5 required languages for data roles — data work is about writing code faster, not running faster code, and ~80% of data work is scraping and preparing data, which Ruby suits perfectly. Argues today's Ruby engineers still wear the data-engineer hat when doing migrations, bulk inserts or normalization, and that data prep often falls into a 'Terra incognita' between backend, devops, QA and data roles. Recommends learning ETL and SQL, mastering n+1 and indexes, and understanding tools like destroy_all vs delete_all. Core warnings: don't be greedy with data, delete unneeded data, avoid 'data-dictated development' (let data serve the app, not the reverse), and treat compliance/security seriously from day one. Case study: a HIPAA-compliant healthcare project where the team inherited PII-laden data from another vendor, couldn't uniquely identify users without consent, was locked into a restrictive compliant hosting provider, and couldn't use BI/analytics/third parties like New Relic or Power BI. After rejecting redesign-from-scratch, they picked data obfuscation (over encryption and tokenization) because it preserves shape and realism. Faker generates fakes but doesn't obfuscate existing production data, so the team built Grazer, a Ruby gem that scans models, emits per-model rule configs, applies Faker-backed strategies that preserve uniqueness and regional relationships (e.g. ZIP-code-consistent addresses), and extracts the obfuscated data as SQL insert statements (full or sliced) — never leaving the server as a dump. A background job tracks the delta to keep obfuscated data consistent as the real data changes.