Paydirt: How I Built an Open-Weight AI Agent to Turn GitHub Bounties into a Real Side Hustle

Description and use

Paydirt turns open-source bounty hunting into a data‑driven side hustle using local Gemma 3 and TabPFN models for payout probability and skill matching, and Temporal workflows for lifecycle tracking.

Problem

Over 60% of open-source bounties never pay out, leaving contributors with uncompensated labor.

Outcome

Paydirt turns bounty hunting into a data-driven side hustle, reducing ghosted issues and enabling contributors to earn by focusing on high-probability payouts.

Workflow

• Discover funded issues by scraping and normalizing bounties across GitHub
• Predict payout probability using TabPFN Bayesian model on 8 tabular features
• Calculate expected value and rank opportunities by hourly earning efficiency
• Match contributor skills locally with Gemma 3 embeddings via Ollama
• Track bounty lifecycle with Temporal workflows and durable timers
• Submit PRs and release bounties with manual confirmation and safety guards

Evidence

Over 60% of open-source bounties never pay out, turning developer side-hustle time into uncompensated labor.

Source

Paydirt turns open-source bounty hunting from a lottery into an evidence-based side hustle.

Source

Uses Prior Labs' TabPFN Bayesian tabular foundation model to score whether a funded issue will actually merge and pay out based on repo velocity, maintainer response intervals, review churn, and bounty size.

Source

Local Skill Matching via Gemma: Embeds your local git commit history and GitHub profile via open-weight Gemma 3 + EmbeddingGemma running entirely on your machine through Ollama.

Source