Đăng Phạm
AI video producer & AI-assisted web tool builder
New to Contra
Đăng is building their profile!
Pulse — Analytics Dashboard
A data-dense SaaS dashboard designed to make heavy metrics feel calm and actionable. I owned the information architecture, visual hierarchy, chart selection, anomaly states, responsive tables, and implementation of the public case study. The resulting system covers 30+ chart states and reduced time-to-insight by 40%.
Case study: https://dangpham.pages.dev/work/analytics-dashboard
0
5
Serenity — Meditation App Redesign
A calming mobile UX concept rebuilt around accessibility, breathing-led navigation, and purposeful micro-interactions. I reduced the home flow to one clear decision, rebuilt the palette to AA+ contrast, designed breath-timed haptics and motion, and created 12 coherent iOS/Android screens.
Case study: https://dangpham.pages.dev/work/meditation-app
0
8
Nova — Dashboard & SaaS UI Kit
A production-ready design system built for real product work: 200+ reusable components, 80+ SaaS screens, tokenized color/type/spacing, complete interaction states, and coordinated light/dark themes. I owned the product strategy, visual system, component architecture, responsive behavior, and case-study implementation.
Case study: https://dangpham.pages.dev/work/nova-ui-kit
0
12
TasteOS — Human-in-the-loop AI Creative Operations
AI video teams can generate more output than they can confidently evaluate. TasteOS makes the missing layer visible: the human judgment that turns synthetic output into work worth shipping.
THE PROTOTYPE
• Command Center — Campaign progress, deadlines, activity, and model usage
• Shot Planner — Six shots across a 12-second sequence
• Variant Lab — A/B/C comparison for identity, motion, texture, and brand fit
• Human Taste Decision Log — Rejected vs approved output, prompt differences, rationale, and measurable results
• Publish Queue — Channel versions protected by a four-part QA gate
THE CORE IDEA
AI generates options. Human taste decides what feels real, on-brand, and ready to publish.
PROCESS
Started with a detailed product prompt containing real content, design tokens, an interaction map, and explicit anti-patterns.
Ran a designer-eye audit for hierarchy, consistency, accessibility, and AI slop.
Refined all five screens without changing the core art direction.
Tested screen switching, variant states, disabled states, and public access in the live prototype.
MCP VERIFICATION
I authenticated Flowstep’s MCP server and used it to verify the TasteOS file, all five screen IDs, the project design guidelines, and the complete prompt and iteration history.
VIEW THE PUBLIC PROTOTYPE
https://app.flowstep.ai/file?activeFileId=9427c575-6fe9-4631-a76e-5f777c0250c6
#FlowstepChallenge #ProductDesign #CreativeAI #AIVideo #MCP
3
180