Impact
2 weeks to 10 minutes
Reduced data onboarding time from 2 weeks to 10 minutes.
80%
Eliminated 80% of manual data engineering work for dataset configuration
3x adoption
Increased data product catalog adoption by 3x in 6 months
5x faster
Enabled teams to ship data products 5x faster with automated profiling
Data teams at enterprise companies were spending weeks manually configuring external datasets — writing schemas, building ingestion pipelines, and validating data quality. This bottleneck prevented them from quickly delivering data products to internal stakeholders.
The platform needed to transform complex data engineering workflows into self-service experiences that non-technical users could complete independently, while maintaining enterprise-grade reliability and governance.
I designed a comprehensive data product creation flow that automates the entire dataset onboarding process:
Smart dataset configuration
Users select source files and the system automatically detects schemas, suggests optimal adapters, and validates data quality without requiring any code.
Progressive disclosure for complexity
Advanced settings (custom adapters, schema versioning, notification rules) are hidden by default but accessible when needed, keeping the core flow simple for 80% of use cases.
Real-time preview and validation
Before finalizing, users see formatted data previews, schema summaries, and matching file counts to catch errors early and build confidence in the configuration.
Guided onboarding with contextual help
Step-by-step checklist shows progress, estimated completion time, and allows users to skip optional steps while ensuring critical configuration is complete.
Key design decisions
Made data browsing feel native
The file explorer uses familiar patterns from macOS Finder and VS Code, with collapsible folders, type indicators, and bulk selection, so data engineers feel immediately comfortable.
Showed data immediately, not abstractions
Rather than showing configuration forms first, users see actual data previews with formatted tables and schema detection, making the impact of their choices tangible.
Built trust through transparency
Every automated decision (adapter selection, schema detection, file matching) is explained with clear reasoning and can be overridden, preventing "black box" anxiety.
Optimized for speed
Default selections, smart recommendations, and skippable optional steps mean power users can create production-ready data products in under 10 minutes.
Outcome
The redesigned data product flow transformed Crux from a developer-focused tool into a self-service platform accessible to product managers and analysts.
Data teams no longer needed dedicated engineering resources for dataset onboarding, freeing engineers to focus on high-value data modeling and pipeline optimization.
The automated profiling and schema detection eliminated entire categories of configuration errors, reducing support tickets by 60% and increasing user confidence in the platform's reliability.
"I want to recognize Alex Dovhyi for his incredible design contributions. He is able to support multiple product managers concurrently. Results of his work are on time but more importantly, deliver high customer value. Despite complex domain his designs are usable and functional. He truly understands the customers needs and product requirements and is able to accommodate our expensive roadmap. He's also a pleasure to work with. Always professional, available. And he has can-do attitude that is so rare to find."
Olga Convey
Sr. Director of Product Management, Crux