Vectors and geometry were one of my most favourite math topics during IIT JEE prep (it's an entrance exam to get into engineering colleges in India) One of the many reasons why I love Qdrant so much!

Did you know that for 1000 requests: - Google offers search API for 5$ but is rate limited to 10k req / day - Bing offered an API for 5$, hiked prices because of AI scrapers, and discontinued it last month - Now Perplexity, Exa, and others are offering it at the same price of 5$

lol I've been using the term "throughtput" my whole life but it's "throughput" (no 't' in the middle) noticed it because of spell correction in vscode while writing a blog.

brb pumping more money into $GOOGL

Modal is absolutely killing it! Congrats Erik Bernhardsson Akshat Bubna!

Q: Do we live in a simulation?

Prioritizing vector search more than keyword search is "the bitter lesson" 101 Embedding models and ANN algorithms improve with more compute and data, while keyword search techniques have remained largely the same for decades.

Very happy to see this! Anthropic has very high talent density. Would be a big win for Indian tech ecosystem if they maintain the same bar for team and projects in India 🇮🇳💪

You'll forget your own code after a few months. Make sure you're keeping it predictable, clean, and documented.

"You don't have everything you want now. But you have everything you wanted 5 years ago!" Let's be grateful :)

Web search is destined evolve into an AI. With each crawl it should update its world knowledge accounting for contradictory beliefs and knowledge. Indexing and storage will fuse into model parameter updates. We will just query the AI. No external database is required.

Yes. It's pronounced Qudrant (not kyu-drant) because it refers to the the vector space 🌌