Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/Praashh/buildml/llms.txt

Use this file to discover all available pages before exploring further.

The Buildml leaderboard is a community ranking page that shows which builders have solved the most AI/ML problems on the platform. Rankings are driven entirely by accepted (PASS) submissions and update automatically as new solutions are judged by the executor.

​
Platform Stats

The leaderboard header surfaces three live community statistics:
StatValue
Builders80+ registered users
Problems20+ available challenges
Submissions500+ total code runs

​
How Ranking Works

A user’s score is the count of distinct problems they have solved with at least one PASS submission. Multiple PASS submissions to the same problem count only once. The user.getLeaderboard tRPC query (public procedure) implements this with a Prisma distinct select: 
const leaderboardData = await ctx.prisma.user.findMany({
  select: {
    id: true,
    name: true,
    image: true,
    submissions: {
      where: { status: "PASS" },
      select: { problemId: true },
      distinct: ["problemId"],   // one entry per unique solved problem
    },
  },
});

const rankedUsers = leaderboardData
  .map((user) => ({
    id: user.id,
    name: user.name ?? "Anonymous",
    image: user.image,
    solvedCount: user.submissions.length,
  }))
  .sort((a, b) => b.solvedCount - a.solvedCount);
The result is an array of { id, name, image, solvedCount } objects sorted in descending order by solvedCount.

​
Leaderboard Table

Each row in the table displays:
  • Rank position — zero-padded integer (e.g., 01, 02). The top three positions are highlighted in the primary accent colour.
  • Avatar — the user’s Google profile image (via NextAuth), with a fallback icon for users without an image.
  • Name — the user’s display name. The currently signed-in user’s row is tinted and labelled You.
  • Solved count — the number of unique problems solved with a PASS.

​
Personal Rank

Signed-in users see a Personal Standing banner above the table showing their current 1-based rank position. This is fetched via the user.getUserRank protected tRPC procedure, which runs the same ranking logic and returns the index of the current user in the sorted array: 
const rank = rankedUsers.findIndex((u) => u.id === currentUserId) + 1;
return rank > 0 ? rank : null;
If the user has not yet solved any problems they will still appear in the list; getUserRank returns null only if the user is not found at all.

​
Authentication

The /leaderboard route is protected by Next.js middleware — unauthenticated visitors are redirected to /signin before they can view the rankings. Once signed in, the Personal Standing banner and the You badge on table rows are shown for the current user. The user.getUserRank query is a protectedProcedure and is only called when a session exists: 
const { data: userRank } = api.user.getUserRank.useQuery(undefined, {
  enabled: !!session,
});

​
Server-Side Prefetching

The leaderboard page uses Next.js server components with tRPC’s HydrateClient pattern to prefetch data before the page is sent to the browser: 
// In LeaderboardPage (server component)
await api.user.getLeaderboard.prefetch();
This means the full ranked list is embedded in the initial HTML payload — no loading spinner is shown on first paint. The client-side LeaderboardClient component hydrates from this cache and re-fetches in the background to keep rankings current.

Build docs developers (and LLMs) love

Get started for freeTalk to us