In a previous post (here) I spoke about starting to mess around with n8n.
Well, a few weeks in things are gaining traction for me here – there are so many small issues I think we can attack at work, and that’s what we’re doing. Starting small to prove out the platform, how we want to host and secure it, define an dev lifecycle for it, etc.
In the last few weeks I’ve played around mostly with an n8n instance on Railway. While there I’ve pulled in a few things – playing with cheap OpenAI models – (looking at you GPT 3.5 Turbo), some Gemini, but a few other things like SerpAPI and Qdrant because of some Udemy labs I was working through.
SerpAPI is a pretty great API aggregator for search APIs. This was particularly helpful with some of the labs I was working on.
One thing that was a bit opaque to me was Qdrant. Over the years (too many to count) – I’ve done a lot with SQL databases (MySQL, SQLServer, Aurora, Oracle, etc), data warehouses (Teradata, Redshift) and even started to shed bias and looked at things like MongoDB and Redis as I’ve started to learn about NoSQL DBs.
However with everything being AI now – these Udemy courses I’ve been taking have led me down a path I didn’t intend to go on. I really wanted to know how these tools work – and I keep poking at the periphery – find a neat topic and dig in.
Enter Vector databases. I knew about them from afar – but found a fantastic short Udemy video (about 4-5 hours depending on how much time you put in) called “Vector Database Fundamentals” taught by Paulo Dichone.
I’ve taken a lot of Udemy courses – and Paulo has to be one of my favorite instructors so far. His presentation, the material, it’s all top notch.
If you don’t know about Vector Databases and have access to Udemy – I’d find the course.
If you don’t have access to Udemy you can still find some of his work on Youtube here
Here is a great 1 hour video to get your interest piqued….
The TLDR is Vector databases allow for a representation of unstructured data in a way that can allow you to do some really interesting work comparing similarity of unstructured data.
This concept on its own will help you really start to understand how things like our friend ChatGPT, Gemini, etc work. It’s well worth the time if you want to know more.