One of the best uses I’ve found for AI is on the projects nobody is paying me to do

2026-05-22 · Source: AI on Medium · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Data Science & Analytics · Depth: Intermediate, quick

Summary

A developer has created a new tool to revive a comprehensive street art event calendar, similar to the defunct Art Crimes calendar from the 1990s. This tool addresses the current challenge of finding event information scattered across Instagram posts. The pipeline involves pulling screenshots of Instagram flyers and captions, then processing them through a vision model to extract structured event data such as date, time, venue, city, artists, and links. This extracted information populates a new calendar hosted at spraystreet.com/events, which the author claims is already the most complete street art calendar available. The project serves both as a passion project and a practical exercise in applying OCR and structured data extraction techniques in a low-stakes environment, demonstrating the value of personal projects for understanding AI model capabilities.

Key takeaway

For AI Engineers or students seeking to truly understand model capabilities, prioritize personal projects over benchmark reviews. If you are struggling to grasp AI's practical limits, build a tool for a problem you genuinely care about. This approach provides invaluable, low-stakes experience in areas like OCR and structured data extraction, revealing what models can and cannot do in real-world scenarios. Your hands-on engagement will accelerate skill development far more effectively.

Key insights

Personal projects using AI for real-world problems offer superior learning and practical skill development compared to benchmarks.

Principles

• Personal projects evaluate AI better than test sets.
• Apply AI to problems you genuinely care about.
• Push AI tools until they break or surprise you.

Method

Screenshots of Instagram posts (flyer + caption) are processed by a vision model to extract structured event data: date, time, venue, city, artists, and links. This data then populates a calendar.

In practice

• Practice OCR and structured data extraction.
• Apply vision models to social media content.
• Build community resources with AI tools.

Topics

• AI Project Development
• Optical Character Recognition
• Structured Data Extraction
• Computer Vision Models
• Instagram Data Processing
• Community Information Systems

Best for: AI Engineer, Software Engineer, AI Student

Related on AIssential

• I Stopped Chasing AI Hype and Started Building Systems That Actually Worked — Focus on practical problem-solving and friction removal to build effective AI systems, not just advanced features.
• [D] Self-Promotion Thread — The community is actively developing diverse open-source tools and frameworks for ML education, research, and practical applications.
• Why You Need Users to Break Your AI Project — Early user interaction is crucial for identifying unforeseen issues and accelerating AI project improvement.
• 8 Real AI Projects Every Developer Should Build — Building practical AI projects is essential for developers to gain real-world skills and career opportunities in 2026.
• 🎙️ This week on How I AI: How to build your own AI developer tools with Claude Code — Custom AI-native dev tools and model-to-model evaluation enhance AI engineering workflows and code quality.
• I Tested 5 OCR Models on 6 Real-World Datasets — Here’s Which One You Should Actually Use — No single OCR model is universally best; performance varies significantly by text type and use case.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI on Medium.