The Problem
Keeping up with the firehose of AI research is a real problem. arXiv alone publishes hundreds of ML papers daily. Add HuggingFace, OpenReview, and Semantic Scholar, and no human can maintain adequate coverage without automated filtering.
Concept
AGI Radar aggregates research papers from multiple sources — arXiv, HuggingFace, OpenReview, and Semantic Scholar — and uses the Gemini API for relevance scoring and prioritization.
How It Would Work
- Multi-source Aggregation — Pull papers from arXiv, HuggingFace model cards and papers, OpenReview submissions, and Semantic Scholar
- Relevance Scoring — Gemini API evaluates each paper against user-defined research interests and priority areas
- Smart Filtering — Configurable thresholds and topic hierarchies to separate signal from noise
- Digest Generation — Daily or weekly summaries of the most relevant papers with AI-generated abstracts
Why Build It
This is a tool I want for my own intellectual workflow. The best side projects solve the builder's own problem first. If it works for me, it likely works for anyone else trying to stay current with ML research without spending hours a day on it.