Skip to Content

UX for AI Data Transformations

Challenge

Design and build an intuitive interface for AI-powered data transformation—an emerging category from ETL that automates cleaning, mapping, and formatting data across complex systems. Existing solutions were either too rigid or too technical, with UIs designed for engineers. Our challenge was to create a UX that welcomed less technical users without losing power or precision.

Role

Head of Design at Osmos.io: I’m on the product team and lead end-to-end UX, working closely with our founders. We’re building fast on the frontier of AI, translating cutting-edge models into clear, usable interfaces. I drive research, stakeholder interviews, interface audits, product design, prototype development, and validation. I work closely with engineers to understand the logic behind AI transformation, then turn that into patterns users can trust. I define interaction models, create detailed mockups in Figma, and lead usability testing to refine clarity and confidence in the system.

Outcomes

  • Helped position Osmos as a leader in the AI ETL category
  • Enabled non-engineers to confidently build data transformations
  • Supported internal workflows and enterprise deployments
  • Much of the work with enterprise clients is confidential and can be discussed up on request.

Press

UX for AI Data Transformation

At Osmos, I lead UX across the product as Head of Design—shaping AI interfaces that make complex data transformation accessible, fast, and enterprise-ready.

Welcome screen for the Osmos AI assistant, featuring an embedded chatbot with guided tutorials and contextual onboarding.

“Our collaboration has successfully addressed challenging issues, and our customers have provided enthusiastic feedback on the improved productivity of Osmos’ intuitive user interface. In our quest for a designer capable of unraveling complex concepts and embracing creative, unconventional solutions, we discovered the perfect fit with Amber.”
—  Naresh Venkat, CTO

Schema editor interface in Osmos for configuring datasets, with field types and validation options for AI-driven data transformation.
Osmos AI value matching interface showing automated field mapping in a data ingestion pipeline UI, optimized for accuracy and ease of use.

AI Value Mapping is one of my inventive & novel solutions, read more about it here ›

When we released our chat companion, we designed flexible artifact cards—multi-modal components that adapt their layout and content based on context, screen size, and use case.

Design system artifacts for Osmos showing compact and expanded UI card states, used to display pipeline data and system activities.
Osmos card UI showcasing various artifacts—pipelines, uploaders, queries, and datasets—organized by frequency and recency.

Planning for contingencies, thinking through corner cases so developers don’t trip over them, iterating fast, and keeping a sense of humor—these are all essential parts of the job.

User safety rail system in Osmos interface showing warning states, confirmation dialogs, tooltips, and conditional flows designed to prevent accidental data loss during AI configuration and validation steps.

Planning through iterations to clear a fast path for the dev team and keep momentum high. Want to know why it’s called Chiñtz-Vicuña?

Osmos notebook builder UI showing lakehouse connection, configuration review, and automated notebook generation steps.
Osmos card UI showcasing various artifacts—pipelines, uploaders, queries, and datasets—organized by frequency and recency.

Built for the future with progressive UIs—this drawer design expands into a full-page experience as we layer in our chat companion.

Instruction editor and resource manager in Osmos Burlap interface, where users add paths and define notebook transformation steps.
Conversational UI for Osmos notebook generation tool, guiding data engineers through creating PySpark-based transformation logic.

Due to confidentiality agreements, only selected portions of my work are shown here. 
I’d be happy to walk you through more details in person or over a call.