Apollo 2.0 — national R&D platform (KISTI)
UI/UX · Design System · 2024

Apollo 2.0 — national R&D platform (KISTI)

UI/UXDesign SystemData

Apollo 2.0 is South Korea's national R&D technology-connection platform, delivered for KISTI (Korea Institute of Science & Technology Information), including views like the 'promising commercialisation' public-R&D forecast. Beyond the screens, I built the admin design system behind it: a full component library so a data-heavy government platform stays consistent and fast to extend.

Role
UI/UX Designer · Design System
Timeline
2024
Team
Astra Technology — for KISTI (government)
Scope
IA, high-fidelity UI & the admin design system
The challenge

Apollo connects two sides that rarely speak the same language: public R&D labs with technology to license, and businesses hunting for the right innovation. As a government-authorised platform centralising national R&D, it had to hold 30,000+ technologies and dense analytics, yet stay scannable for expert users and credible to government stakeholders. The hard part wasn't search. It was matchmaking.

FigmaDesign SystemInformation ArchitectureData Viz
Approach
  1. 01

    Plan

    Scoped the platform with the KISTI team and agreed what 'connection' meant in practice.

  2. 02

    Architect

    Built the information architecture to make heavy research data quick to scan.

  3. 03

    Design

    Delivered high-fidelity UI tuned for expert users.

The battles
A national database that opens with a question
Battle 01 · The question

A national database that opens with a question

Apollo holds 30,000+ public R&D technologies, and the obvious design is a search bar on top of a database. But the brief was never search. It was matchmaking between labs that have technology and businesses that need it. So the explore screen opens like a conversation: name the item you care about, and the platform recommends the closest global items, with inline tooltips walking a first-time user through what each menu actually forecasts. A search engine assumes you know what to look for; a matchmaker only asks what you have. That reframing is what pushed successful connections up 65%.

One page that reads like a briefing, not a dump
Battle 02 · Density

One page that reads like a briefing, not a dump

A single item page carries a plain-language description, a five-year indicator trend, an item-centred network graph and a related-items table, the kind of load that usually collapses into clutter. The rule was density with breathing room, sequenced the way an analyst thinks: read what it is, see where it's heading, see what it connects to, then compare the neighbours. Scores wear colour-graded chips so a 99 and a 32 register before you've read the digits, and bookmark-and-compare sits on every row because reading was never the end goal. Density, here, is respect for the expert reader.

A score means nothing until it stands next to something
Battle 03 · The verdict

A score means nothing until it stands next to something

Corporate Prospect Analysis rates thousands of companies, but a raw score doesn't help anyone decide. So every number ships with its context: a sunburst slices the Top 100 by Korea's industrial classification, letter grades turn seven metrics into a report card you can scan down a table, and each company card sets its bars against the regional and industry averages, with a radar showing the shape of its strengths. The same Highcharts language runs through all of it. The design bet: experts don't want the system's answer. They want the evidence arranged so their own answer comes fast.

Government platforms outlive their designers
Battle 04 · The system underneath

Government platforms outlive their designers

The screens are the visible half; the Apollo Admin design system is the half that keeps them honest. I organised it into four pinned families (Input, Display, Feedback, Navigation) plus an icon set and Highcharts-based data viz, with the table spec drawn down to every cell variant, status chip and header state. On a data-heavy platform a table isn't one component, it's the product, so it earned the deepest page in the library. The point of the system is time: screen forty has to look like it was designed on the same day as screen four, even when it's built years later by someone else.

Not a search engine — a matchmaker.

Key decisions

Density with breathing room

Expert users want to see a lot at once, the layout shows depth without tipping into clutter.

Outcome
  • ✦65% increase in successful technology connections
  • ✦92% satisfaction among government stakeholders
  • ✦30,000+ technologies actively maintained
  • ✦Recognised as a cornerstone of South Korea's innovation infrastructure
Reflection

Designing at national scale taught me to respect density without fearing it: expert users want to see a lot, fast. The real win wasn't a prettier screen; it was turning plain search into guided matchmaking, which is what actually moved successful connections up 65%.

Next project

Gobiz Vietnam — B2B trade marketplace

Like what you see? Let's talk.