community-driven task pool

Contribute a Task

The task pool is small and we need your help growing it. Community contributions are the core moat. Any problem where code is "tedious, time-consuming, or hitting performance bottlenecks" is a perfect candidate. Fill out the form below and our team will review it.

// three tiers

Choose Your Level

Contribute at whatever depth matches your time and resources. Every tier helps grow the benchmark pool.

industrial benchmark

High Tier

Experienced developers with a full benchmark ready to go.

includes
  • ▸task.toml
  • ▸instruction.md
  • ▸environment/ (base code)
  • ▸tests/ (automated scoring)
  • ▸solution/ (optional reference)
fast_inv_sqrt/
├── task.toml
├── instruction.md
├── environment/
│   ├── Makefile
│   └── main.c
├── tests/
│   ├── test_runner.py
│   └── test_cases.json
└── solution/
    └── fast_main.c
after submission

Directly adopted into Open Research sandbox as a live benchmark.

reward

Core Author credit on the project site.

skeleton frame

Mid Tier

Developers who have unoptimized code but no time to build the full test suite.

includes
  • ▸instruction.md
  • ▸environment/ (base code)
fast_inv_sqrt_mid/
├── instruction.md
└── environment/
    └── main.c
after submission

AutoLab team builds the tests, task.toml, and scoring pipeline for you.

reward

Contributor credit on the project site.

concept phase

Low Tier

Anyone with an optimization pain point. No code required.

includes
  • ▸Text description of the challenge
No folder structure needed.
Just describe the bottleneck
and why it matters.
after submission

Enters the candidate pool. If high-value, the team builds it from scratch.

reward

Listed in the candidate pool with community upvotes.

// submit

Propose a Task

Select a tier and fill in the details. Your proposal is submitted privately — the team will review and follow up via email. Approved proposals appear in the candidate pool.

select tier
private — only for team follow-up
Reviewed privately by the AutoLab team.