Ruohan Li

TikTok Search · In progress 2026

Making AI search feel more natural

Original TikTok Search project overview
Product design work across AI-generated search results and core search architecture.
Role
Product Design
Intern
Timeline
June 2026
Present
Focus
Search AI
Core architecture
Skills
UX/UI · AIGC
Systems · Motion

The answers had the right information, but not always the right rhythm.

The experience needed a clearer relationship between text, images, video, and the space that holds them together. Inconsistent media patterns made rich answers harder to scan than they needed to be.

Media design optimization for AI-generated search results Second AI-generated search result design direction

NDA note. The redesigned screens have not launched, so this case study shares my process and design principles without showing confidential final work.

I mapped the system before redesigning the surface.

Instead of jumping into polished screens, I first built a view of the existing components, content types, edge cases, and competing interaction patterns.

What surfaced

Low presentation efficiency, unused component patterns, and a weak media placement strategy.

01

Audit

Documented current patterns and the places where they broke down.

02

Benchmark

Compared major products to understand display efficiency and content-media coordination.

03

Frame

Grouped the problem around media consistency, whitespace, and interaction behavior.

Design direction

One media language, less dead space, clearer trade-offs.

The direction focused on unifying image and video containers while preserving enough breathing room for dense AI answers.

Exploration 01

Horizontal scroll

Useful for preserving vertical space and showing that more media is available.

Exploration 02

Full width

Stronger visual emphasis and simpler scanning, with a larger vertical footprint.

Preparing to validate

I isolated variables such as container size and badge placement so the team could review and test specific decisions.

More screen does not automatically mean more clarity.

I owned TikTok Core Search iOS foldable-screen adaptation end to end, turning an ambiguous scope into a clear framework for active search and passive discovery. I identified edge cases that standard rules could not cover and proposed a systematic classification that gave engineering a clear implementation spec.

Original foldable phone adaptation project summary
Original project summary covering search architecture, split-screen behavior, special scenarios, and design handoff.
Full Figma workspace for TikTok foldable-screen search adaptation
Full working file spanning responsive breakpoints, search scenarios, component rules, and design handoff.
Delivery

I delivered Figma specifications across multiple screen widths and a structured PRD, aligning design, product, and engineering with minimal back-and-forth.

Mode A

Active search

Mode B

Passive discovery

This work is still in progress. Please email or Lark me for more details if you’re internal.

Ruohan standing beneath a glowing TikTok logo

Simplicity came from understanding the system.

The visible interface was only the final layer. The real work was organizing component behavior, media rules, platform constraints, and edge cases into decisions the team could evaluate.

Next project ByteDance AI