Paydirt: How I Built an Open-Weight AI Agent to Turn GitHub Bounties into a Real Side Hustle
What I Built : Every open-source contributor knows the sting of bounty ghosting: you spend two evenings fixing a thorny bug on a funded GitHub issue only to find the maintainer hasn't logged in since August, another contributor opened an unreviewed PR three weeks prior, or the bounty platform escrow terms are predatory. Over 60% of open-source bounties never pay out, turning developer side-hustle…
Paydirt is an open-weight AI agent designed to turn GitHub bounty hunting into a profitable side hustle. Developers often face issues like "bounty ghosting" where maintainers fail to respond or update bounties, resulting in a 60% payout failure rate. Paydirt addresses this by providing evidence-based predictions of payout probability for funded issues on GitHub.
The AI agent discovers funded issues across GitHub, predicts the real payout probability using Prior Labs' TabPFN Bayesian tabular foundation model, and calculates the expected value by ranking opportunities based on hourly earning efficiency. It matches local skills using Gemma 3, an open-weight model, embedding developer git commit history and GitHub profiles without sharing data with third-party cloud APIs.
Temporal, a workflow management system, manages the end-to-end lifecycle of bounties, from claiming intent to PR submission and bounty release. Built-in safety measures prevent accidental spam and require manual confirmation for external GitHub mutations. The platform is designed for freelancers, students, and independent OSS maintainers who can dedicate 10-20 hours a week towards monetizing bounties without wasting time on ghosted issues.
The monthly subscription or 2% success fee on settled payouts ensures a clear return on investment within a single saved weekend. The platform is available at https://paydirt-iota.vercel.app/ and its code repository is located at https://github.com/subhranshudash13-dotcom/Paydirt. The monorepo structure includes shared types, bounty ranking algorithms, backend, worker workflows, web applications, and a predictor sidecar for TabPFN.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
