HackerTube: Building a Curated YouTube Alternative via AI
A developer leverages LLMs to filter Hacker News recommendations into a clean video platform.
1Algorithmic Fatigue and Motivation
The developer initiated this project after experiencing frustration with YouTube's recommendation engine, which they described as an enshittified wall of slop. Seeking a more intellectual viewing experience, they decided to bypass mainstream algorithms entirely. The goal was to build a platform that exclusively features content recommended by the Hacker News community since 2007.
2Leveraging Opus 5.5 Capabilities
The entire development process was accelerated by using Opus 5.5 to handle complex tasks. The AI was tasked with writing a YouTube clone from scratch, including the necessary infrastructure and logic. This automation allowed the creator to move from a conceptual frustration on Friday night to a functional live website by Saturday morning.
3Data Processing and Filtering
The pipeline involved downloading all historical Hacker News comments to identify relevant video links. By running Sonnet over the dataset, the system successfully filtered out spam and low-quality content. This rigorous curation process narrowed down the massive volume of discussions into a refined list of 40,000 comments that specifically recommended YouTube videos.
4Content Scope and Scale
The final database comprises 6,454 unique YouTube videos sourced from community recommendations spanning the years 2007 to 2026. These videos are categorized across 150 distinct topics, ensuring a diverse range of technical and intellectual content. This dataset represents a human-curated alternative to the standard, engagement-driven recommendations found on the primary YouTube interface.
5Infrastructure and Deployment
After processing the data, the developer wrapped the content in a custom YouTube-style user interface. The application was then deployed to a virtual private server, where it underwent hardening and optimization processes. The entire technical stack was built and secured in approximately two hours, demonstrating the efficiency of modern AI-assisted software development workflows.
6Technical Hardening and Security
Beyond the basic functionality, the developer prioritized system security and performance optimization during the deployment phase. By hardening the VPS, they ensured the platform could handle traffic while maintaining a clean, minimalist environment. This focus on backend stability allows the site to remain functional without the overhead of traditional, ad-heavy video hosting platforms.
7Platform Accessibility and Domain
The project is now live at hackertube.org, serving as a functional repository for high-signal video content. By centralizing links that have been vetted by the Hacker News community, the site provides a unique discovery mechanism. It effectively transforms scattered forum discussions into a structured, searchable video library that avoids the pitfalls of mainstream algorithms.
8Implications for Content Discovery
This project highlights a growing trend of developers building bespoke tools to reclaim control over their digital consumption. By combining LLM-driven data processing with community-sourced curation, the creator successfully bypassed the limitations of mass-market recommendation engines. It serves as a practical example of how AI can rapidly bridge the gap between complex data and user-friendly interfaces.