AI resources

Below, you’ll find the AI resources I maintain.

"I've been really enjoying your podcast, Best AI papers explained, for the last few months. Great choice of papers, and it is interesting how AI boosts researchers' productivity and helps us digest information." 

- From a computer science professor/researcher in a tech firm research lab

“Best AI Paper Explained” began as my personal daily effort to stay on top of cutting-edge AI research. It has since grown into a bi-daily podcast with over 500 subscribers and more than 150 daily listeners. I curate the newest and most important research papers you shouldn’t miss. 

I engineered a hybrid RAG architecture designed for high-performance retrieval without the high cost of commercial APIs. The frontend is hosted on Hugging Face Spaces (Streamlit), while the heavy inference runs locally on my server using Ollama (Qwen3-8B), bridged by a secure Zrok tunnel.

Beyond the architecture, I improved response quality by implementing Hybrid Search, combining vector embeddings with keyword matching. This ensures the bot captures both broad concepts and specific technical acronyms. I also used Semantic Chunking to retain research ideas during retrieval and Contextual Memory to enable the bot to understand follow-up questions naturally.