eda-bench – Data Exploration & Reasoning Benchmark

Overview

Research-grade benchmarking site for evaluating how effectively large language models explore and reason about data, with a live leaderboard, task registry, and documentation.

Problem

The project needed a public home that could present complex benchmark results in a way that researchers and practitioners could actually use.

Approach

I built a Next.js 16 App Router site with an animated leaderboard, GitHub-backed task registry, and MDX-powered docs.

Key features

• Live leaderboard with smooth GSAP-powered transitions
• Task registry that is filterable and sourced from GitHub
• Integrated docs using MDX
• One-click copy for plain text and BibTeX citations

Outcomes

• Leaderboard, tasks, docs, and contributors all live under one consistent design system
• Built to support papers, talks, and community contributions