Redo
Senior product designer
At Redo, I joined as the second designer and am currently leading design initiatives. My focus has been on taking multiple 0-to-1 products from concept to launch while also spearheading the development of our internal design system. I’ve played a key role in shaping Redo’s design foundation and supporting our brand as we continue to scale.
Design system
The problem
Redo was scaling multiple 0-to-1 products (Support, Order Management, Reviews, Order tracking, Marketing, Warehouse management, and Inventory management) with a design team of 3. Without a shared visual foundation, every new surface risked drifting in type, color, spacing, and component usage. We risked slowing shipping and eroding consistency.
The craft decision
I built Redo's token architecture based on the Tailwind structure as primitive tokens → semantic tokens → component-level tokens. Rather than optimizing purely for Figma-side reuse, I designed the token structure so it could be read and extended by AI coding tools. With designers contributing code on a daily basis AI-assisted code is top of mind.
The component side is a much more interesting problem. Shadcn is the leader in the space as far as vibe-coded accuracy goes but Redo’s product is too broad to warrant swapping now. We are optimizing our current component library that we built to track prop usage and AI accuracy.
Validating in code + the craft bar
We are building a bespoke design system documentation site that can be published for designers, engineers, and most importantly AI to consume. We are also building dashboards to track usage metrics for tokens, components, and props.
Outcome + next steps
The goal is to use these tools to increase quality and coverage. We have baseline metrics to measure against. We are tracking first-shot vibe code accuracy via quantitative analysis of component and token usage and also qualitative analysis via automated Playwright spot checks.
I am also looking to trim the token set down and reduce our semantic layer to only allow for component-level tokens. No more `utility-success-600`. We are instead enforcing `button-success-bg`. This will better force designers to be deliberate about the colors they are choosing and ensuring that the primitive layer is never reflected in the front-end.
Support
Redesigned the Support interface from the ground up to deliver better visibility into cross-product issues across the Redo platform and enable future AI-first support experiences for both web and mobile.
- Overhauled the entire UI to streamline workflows and highlight actionable trends across products.
- Shaped data visualization and interaction patterns for smarter, faster support resolutions.
- Collaborated with engineering to lay the groundwork for AI-driven troubleshooting and predictive support.
The result set the foundation for more integrated, proactive, and intelligent support tools for Redo customers.
Order management
Redesigned the Support interface from the ground up to deliver better visibility into cross-product issues across the Redo platform and enable future AI-first support experiences for both web and mobile.
- Overhauled the entire UI to streamline workflows and highlight actionable trends across products.
- Shaped data visualization and interaction patterns for smarter, faster support resolutions.
- Collaborated with engineering to lay the groundwork for AI-driven troubleshooting and predictive support.
The result set the foundation for more integrated, proactive, and intelligent support tools for Redo customers.
Redo
Senior product designer
At Redo, I joined as the second designer and am currently leading design initiatives. My focus has been on taking multiple 0-to-1 products from concept to launch while also spearheading the development of our internal design system. I’ve played a key role in shaping Redo’s design foundation and supporting our brand as we continue to scale.
Design system
The problem
Redo was scaling multiple 0-to-1 products (Support, Order Management, Reviews, Order tracking, Marketing, Warehouse management, and Inventory management) with a design team of 3. Without a shared visual foundation, every new surface risked drifting in type, color, spacing, and component usage. We risked slowing shipping and eroding consistency.
The craft decision
I built Redo's token architecture based on the Tailwind structure as primitive tokens → semantic tokens → component-level tokens. Rather than optimizing purely for Figma-side reuse, I designed the token structure so it could be read and extended by AI coding tools. With designers contributing code on a daily basis AI-assisted code is top of mind.
The component side is a much more interesting problem. Shadcn is the leader in the space as far as vibe-coded accuracy goes but Redo’s product is too broad to warrant swapping now. We are optimizing our current component library that we built to track prop usage and AI accuracy.
Validating in code + the craft bar
We are building a bespoke design system documentation site that can be published for designers, engineers, and most importantly AI to consume. We are also building dashboards to track usage metrics for tokens, components, and props.
Outcome + next steps
The goal is to use these tools to increase quality and coverage. We have baseline metrics to measure against. We are tracking first-shot vibe code accuracy via quantitative analysis of component and token usage and also qualitative analysis via automated Playwright spot checks.
I am also looking to trim the token set down and reduce our semantic layer to only allow for component-level tokens. No more `utility-success-600`. We are instead enforcing `button-success-bg`. This will better force designers to be deliberate about the colors they are choosing and ensuring that the primitive layer is never reflected in the front-end.
Support
Redesigned the Support interface from the ground up to deliver better visibility into cross-product issues across the Redo platform and enable future AI-first support experiences for both web and mobile.
- Overhauled the entire UI to streamline workflows and highlight actionable trends across products.
- Shaped data visualization and interaction patterns for smarter, faster support resolutions.
- Collaborated with engineering to lay the groundwork for AI-driven troubleshooting and predictive support.
The result set the foundation for more integrated, proactive, and intelligent support tools for Redo customers.
Order management
Redesigned the Support interface from the ground up to deliver better visibility into cross-product issues across the Redo platform and enable future AI-first support experiences for both web and mobile.
- Overhauled the entire UI to streamline workflows and highlight actionable trends across products.
- Shaped data visualization and interaction patterns for smarter, faster support resolutions.
- Collaborated with engineering to lay the groundwork for AI-driven troubleshooting and predictive support.
The result set the foundation for more integrated, proactive, and intelligent support tools for Redo customers.
Redo
Senior product designer
At Redo, I joined as the second designer and am currently leading design initiatives. My focus has been on taking multiple 0-to-1 products from concept to launch while also spearheading the development of our internal design system. I’ve played a key role in shaping Redo’s design foundation and supporting our brand as we continue to scale.
Design system
The problem
Redo was scaling multiple 0-to-1 products (Support, Order Management, Reviews, Order tracking, Marketing, Warehouse management, and Inventory management) with a design team of 3. Without a shared visual foundation, every new surface risked drifting in type, color, spacing, and component usage. We risked slowing shipping and eroding consistency.
The craft decision
I built Redo's token architecture based on the Tailwind structure as primitive tokens → semantic tokens → component-level tokens. Rather than optimizing purely for Figma-side reuse, I designed the token structure so it could be read and extended by AI coding tools. With designers contributing code on a daily basis AI-assisted code is top of mind.
The component side is a much more interesting problem. Shadcn is the leader in the space as far as vibe-coded accuracy goes but Redo’s product is too broad to warrant swapping now. We are optimizing our current component library that we built to track prop usage and AI accuracy.
Validating in code + the craft bar
We are building a bespoke design system documentation site that can be published for designers, engineers, and most importantly AI to consume. We are also building dashboards to track usage metrics for tokens, components, and props.
Outcome + next steps
The goal is to use these tools to increase quality and coverage. We have baseline metrics to measure against. We are tracking first-shot vibe code accuracy via quantitative analysis of component and token usage and also qualitative analysis via automated Playwright spot checks.
I am also looking to trim the token set down and reduce our semantic layer to only allow for component-level tokens. No more `utility-success-600`. We are instead enforcing `button-success-bg`. This will better force designers to be deliberate about the colors they are choosing and ensuring that the primitive layer is never reflected in the front-end.
Support
Redesigned the Support interface from the ground up to deliver better visibility into cross-product issues across the Redo platform and enable future AI-first support experiences for both web and mobile.
- Overhauled the entire UI to streamline workflows and highlight actionable trends across products.
- Shaped data visualization and interaction patterns for smarter, faster support resolutions.
- Collaborated with engineering to lay the groundwork for AI-driven troubleshooting and predictive support.
The result set the foundation for more integrated, proactive, and intelligent support tools for Redo customers.
Order management
Designed and launched the first version of Redo’s OMS, bridging modern software with real-world warehouse practices.
- Created an intuitive system that syncs warehouse processes with AI-powered automation and recommendations.
- Focused on features that boost speed, accuracy, and efficiency for warehouse workers—outpacing traditional OMS products.
- Developed workflows that adapt to physical constraints and empower teams with actionable insights.
The OMS quickly improved operational efficiency, reduced errors, and delivered value to customers seeking best-in-class automation.