> program: databounty_open · contribution: verified▋
Build the open corpus.
Earn recognition.
Contribute to carefully specified coding datasets. Every accepted item is checked for quality, provenance, and task-specific evidence before it is credited.
✓ dedupe/✓ contamination/✓ sandboxed tests/✓ llm review➜ accepted work earns credit
open
contribution programs
evidence
quality before credit
attribution
recognition for accepted work
// dataset_specs
Browse the work programs currently described on DataBounty. Availability and rewards are confirmed in the dashboard.
debugging · ts/react● active
React Hooks Debugging Dataset
4,500USDC · 6/item
312 / 50062%
audit 25%08-15view →
error_diag · py/asyncio● active
Python Async Error Diagnosis Corpus
9,800USDC · 5.50/item
486 / 1,20041%
exclusive09-01view →
migration · ts/prisma● active
SQL to Prisma Migration Pairs
5,400USDC · 5/item
174 / 80022%
non-exclusive08-30view →
test_gen · go/testing● active
Go Table-Driven Test Generation Set
13,200USDC · 4.50/item
1,231 / 2,00062%
audit 25%10-01view →
function · rust/std● pilot_review
Rust Trait Implementation Challenges
11,250USDC · 11/item
58 / 600pilot 6/6
full audit09-20view →
debugging · ts/vue3● pilot_open
Vue 3 Reactivity Bug Dataset
3,400USDC · 6/item
0 / 400pilot 7/20
audit 25%09-10view →
// verification_not_vibes
Every accepted item earned its way through an automated pipeline, payment only follows acceptance.
21,840
exec-verified items
Broken code must fail the tests, fixed code must pass, in a sandbox, every time.
11%
dup rejection rate
Near-duplicates are caught by similarity scoring before they reach validation.
52,410
contamination checks
Every item is screened against public benchmarks so datasets don't leak eval sets.
audits
independent review
Ranked validators spot-check accepted batches and earn bonuses for confirmed issues.
// how_the_loop_works
From spec to delivered dataset, every step is funded, verified, and auditable.
01 create_bounty → 02 llm_plan → 03 fund_pilot_tranche → 04 pilot_review (7-day auto-approve)
→ 05 fund_full_tranche → 06 creators_claim_batches → 07 validation_pipeline
→ 08 independent_audits → 09 accepted_work_paid → 10 dataset_delivered ▋
// built_for_all_three_sides
Requesters fund, creators build, validators keep everyone honest.
requesters
Pay only for accepted work
✓Pilot-first funding: ~1–2% of budget at risk before you see real samples.
✓Escrowed USDC released only when items pass validation and audit.
✓Choose exclusive or non-exclusive license per bounty.
creators
Claim batches, ship items, get paid
✓Claim 10-item batches with clear specs and per-item USDC rewards.
✓AI assistance is allowed, disclose your method, we verify the output.
✓No bond while you deliver on time; higher ranks unlock bigger batches.
validators
Audit batches, earn bonuses
✓Review batches of provisionally accepted items against the spec.
✓Base reward per batch plus a bonus for every confirmed issue.
✓Auditor ranks unlock larger batches and bonus multipliers.
// recently_delivered
Completed datasets with public quality stats and payout records.
function/feature● delivered
TypeScript Generics Implementation Set
accepted_items1,000
total_paid6,240 USDC
licensenon-exclusive
delivered2026-05-04
debugging● delivered
React Re-render Debugging Pack
accepted_items600
total_paid7,800 USDC
licenseexclusive
delivered2026-04-19
migration◐ partial
Pandas Migration to Polars Corpus
accepted_items512
total_paid3,286 USDC
licensenon-exclusive
delivered2026-04-02
databounty open
Build trusted coding data,
one verified contribution at a time.
Whether you are requesting a dataset or contributing to one, the workflow records what was submitted, checked, and accepted.
