2023
ML
Image Search
New Service Initiative
New ML service
Discover how machine learning and UX came together to transform the way of searching — faster, smarter, and tailored to real target user behavior.
TL;DR
TL;DR
Designed and led the development of an ML-powered image search app that helps climbers find bouldering solutions faster through photo-based discovery and personalized filters.
Responsible for
Led a cross-functional team of 5, managing both product design and project timelines using agile methodology
Designed the full app experience from scratch, including structure, interaction, visual system, micro-interaction, and brand identity
Conducted user research and usability testing with over 150 climbers, integrating insights into final UX decisions
Tools used
Figma
Protopie
Adobe Illustrator
Adobe After Effect
Lottie
Final Cut Pro & Motion
Problem
Problem
Climbers were frustrated by the lack of fast, reliable ways to find solutions when stuck on bouldering problems.
Since climbing problems rely on color-coded holds and spatial patterns, text-based search isn't ideal. However, most climbers still rely on Instagram or YouTube to find bouldering hints-often spending too much time scrolling through unrelated content and struggling to identify the exact problem from small thumbnails. These platforms also lack filters for physical traits like height or reach, which are essential for personalized recommendations.
To solve this, my team wanted to build a smarter, image-based search experience powered by machine learning — designed to help climbers get "unstuck" faster and more effectively.
Design Process
Design Process
From hypothesis to high-fidelity prototype, I validated real user needs, iterated designs, and aligned with developers through agile collaboration.
I began by surveying 100 climbers to test our hypothesis — 88% searched for bouldering hints. Using an Agile Team Charter, I defined feature-level tasks and collaborated with the development team to estimate workloads across front-end, back-end, and interface areas.
To stay aligned, I created quick paper prototypes before moving into low-fidelity design. After conducting A/B tests with internal team members, I iterated on the concepts and built high-fidelity prototypes with visuals and micro-interactions.
With a working prototype in hand, I conducted usability testing with 50 climbers — from beginners to professionals — in actual indoor climbing gym settings. The real-world feedback led to key design refinements that shaped the final product.
Final Design
Final Design
The final design delivered an innovative image-based search experience tailored to real climbers' behavior and gym routines.
The app allows users to search for bouldering problems by simply taking a photo, with machine learning matching visual patterns to video content. Filters for height and reach help personalize search results, while the video seekbar enables users to replay specific movements easily.
Visual clarity and micro-interactions were refined through multiple usability iterations, ensuring a seamless experience for climbers of all levels.
Impact
Impact
92% of users responded that image-based search was revolutionary and made it easier to find bouldering problems compared to existing methods.
The final design addressed key frustrations from traditional platforms through photo-based search and personalized filters.
Lesson learned
Lesson learned
This project taught me how to design intuitive ML-powered experiences while leading agile development in a cross-functional team.
Designing for Machine Learning UX
This project gave me hands-on experience designing a UX flow that integrated machine learning in a meaningful, user-facing way. I learned how to surface ML outputs—like visual pattern recognition—in ways that felt intuitive, useful, and trustworthy to users. It reinforced the importance of simplifying complexity and aligning intelligent features with real user intent.
Practicing Agile Leadership in a Cross-Functional Team
As the project manager, I guided a team unfamiliar with agile through iterative sprints. Visualizing roadmaps and aligning timelines taught me how to prioritize effectively and foster collaborative momentum across design and development.
Interested in the full story?
I'd love to share more behind-the-scenes insights - feel free to reach out for a deeper dive.
More Works
(NY® — 02)
©2024
More Works
(NY® — 02)
©2024
2023
ML
Image Search
New Service Initiative
New ML service
Discover how machine learning and UX came together to transform the way of searching — faster, smarter, and tailored to real target user behavior.
TL;DR
Designed and led the development of an ML-powered image search app that helps climbers find bouldering solutions faster through photo-based discovery and personalized filters.
Responsible for
Led a cross-functional team of 5, managing both product design and project timelines using agile methodology
Designed the full app experience from scratch, including structure, interaction, visual system, micro-interaction, and brand identity
Conducted user research and usability testing with over 150 climbers, integrating insights into final UX decisions
Tools used
Figma
Protopie
Adobe Illustrator
Adobe After Effect
Lottie
Final Cut Pro & Motion
Problem
Climbers were frustrated by the lack of fast, reliable ways to find solutions when stuck on bouldering problems.
Since climbing problems rely on color-coded holds and spatial patterns, text-based search isn't ideal. However, most climbers still rely on Instagram or YouTube to find bouldering hints-often spending too much time scrolling through unrelated content and struggling to identify the exact problem from small thumbnails. These platforms also lack filters for physical traits like height or reach, which are essential for personalized recommendations.
To solve this, my team wanted to build a smarter, image-based search experience powered by machine learning — designed to help climbers get "unstuck" faster and more effectively.
Design Process
From hypothesis to high-fidelity prototype, I validated real user needs, iterated designs, and aligned with developers through agile collaboration.
I began by surveying 100 climbers to test our hypothesis — 88% searched for bouldering hints. Using an Agile Team Charter, I defined feature-level tasks and collaborated with the development team to estimate workloads across front-end, back-end, and interface areas.
To stay aligned, I created quick paper prototypes before moving into low-fidelity design. After conducting A/B tests with internal team members, I iterated on the concepts and built high-fidelity prototypes with visuals and micro-interactions.
With a working prototype in hand, I conducted usability testing with 50 climbers — from beginners to professionals — in actual indoor climbing gym settings. The real-world feedback led to key design refinements that shaped the final product.
Final Design
The final design delivered an innovative image-based search experience tailored to real climbers' behavior and gym routines.
The app allows users to search for bouldering problems by simply taking a photo, with machine learning matching visual patterns to video content. Filters for height and reach help personalize search results, while the video seekbar enables users to replay specific movements easily.
Visual clarity and micro-interactions were refined through multiple usability iterations, ensuring a seamless experience for climbers of all levels.
Impact
92% of users responded that image-based search was revolutionary and made it easier to find bouldering problems compared to existing methods.
The final design addressed key frustrations from traditional platforms through photo-based search and personalized filters.
Lesson learned
This project taught me how to design intuitive ML-powered experiences while leading agile development in a cross-functional team.
Designing for Machine Learning UX
This project gave me hands-on experience designing a UX flow that integrated machine learning in a meaningful, user-facing way. I learned how to surface ML outputs—like visual pattern recognition—in ways that felt intuitive, useful, and trustworthy to users. It reinforced the importance of simplifying complexity and aligning intelligent features with real user intent.
Practicing Agile Leadership in a Cross-Functional Team
As the project manager, I guided a team unfamiliar with agile through iterative sprints. Visualizing roadmaps and aligning timelines taught me how to prioritize effectively and foster collaborative momentum across design and development.
Interested in the full story?
I'd love to share more behind-the-scenes insights - feel free to reach out for a deeper dive.
More Works
(NY® — 02)
©2024
2023
ML
Image Search
New Service Initiative
New ML service
Discover how machine learning and UX came together to transform the way of searching — faster, smarter, and tailored to real target user behavior.
TL;DR
Designed and led the development of an ML-powered image search app that helps climbers find bouldering solutions faster through photo-based discovery and personalized filters.
Responsible for
Led a cross-functional team of 5, managing both product design and project timelines using agile methodology
Designed the full app experience from scratch, including structure, interaction, visual system, micro-interaction, and brand identity
Conducted user research and usability testing with over 150 climbers, integrating insights into final UX decisions
Tools used
Figma
Protopie
Adobe Illustrator
Adobe After Effect
Lottie
Final Cut Pro & Motion
Problem
Climbers were frustrated by the lack of fast, reliable ways to find solutions when stuck on bouldering problems.
Since climbing problems rely on color-coded holds and spatial patterns, text-based search isn't ideal. However, most climbers still rely on Instagram or YouTube to find bouldering hints-often spending too much time scrolling through unrelated content and struggling to identify the exact problem from small thumbnails. These platforms also lack filters for physical traits like height or reach, which are essential for personalized recommendations.
To solve this, my team wanted to build a smarter, image-based search experience powered by machine learning — designed to help climbers get "unstuck" faster and more effectively.
Design Process
From hypothesis to high-fidelity prototype, I validated real user needs, iterated designs, and aligned with developers through agile collaboration.
I began by surveying 100 climbers to test our hypothesis — 88% searched for bouldering hints. Using an Agile Team Charter, I defined feature-level tasks and collaborated with the development team to estimate workloads across front-end, back-end, and interface areas.
To stay aligned, I created quick paper prototypes before moving into low-fidelity design. After conducting A/B tests with internal team members, I iterated on the concepts and built high-fidelity prototypes with visuals and micro-interactions.
With a working prototype in hand, I conducted usability testing with 50 climbers — from beginners to professionals — in actual indoor climbing gym settings. The real-world feedback led to key design refinements that shaped the final product.
Final Design
The final design delivered an innovative image-based search experience tailored to real climbers' behavior and gym routines.
The app allows users to search for bouldering problems by simply taking a photo, with machine learning matching visual patterns to video content. Filters for height and reach help personalize search results, while the video seekbar enables users to replay specific movements easily.
Visual clarity and micro-interactions were refined through multiple usability iterations, ensuring a seamless experience for climbers of all levels.
Impact
92% of users responded that image-based search was revolutionary and made it easier to find bouldering problems compared to existing methods.
The final design addressed key frustrations from traditional platforms through photo-based search and personalized filters.
Lesson learned
This project taught me how to design intuitive ML-powered experiences while leading agile development in a cross-functional team.
Designing for Machine Learning UX
This project gave me hands-on experience designing a UX flow that integrated machine learning in a meaningful, user-facing way. I learned how to surface ML outputs—like visual pattern recognition—in ways that felt intuitive, useful, and trustworthy to users. It reinforced the importance of simplifying complexity and aligning intelligent features with real user intent.
Practicing Agile Leadership in a Cross-Functional Team
As the project manager, I guided a team unfamiliar with agile through iterative sprints. Visualizing roadmaps and aligning timelines taught me how to prioritize effectively and foster collaborative momentum across design and development.
Interested in the full story?
I'd love to share more behind-the-scenes insights - feel free to reach out for a deeper dive.
More Works
©2024