Gaurav Sen
Resources:
Neon Vector Database: https://fyi.neon.tech/gs1
AWS Transcribe: https://aws.amazon.com/pm/transcribe
This is how I built an AI teaching assistant with vector databases and ChatGPT. The bot uses the RAG model to answer user questions in real-time.
This project took some time to create, and you can see it in action below:
https://interviewready.io/learn/system-design-course
00:00 Agenda
00:16 Problem Statement
01:15 Vanilla ChatGPT
03:22 Vector Databases
06:49 Implementation
07:42 Internal Algorithms
09:30 Demo
11:17 RAG Model
12:03 Hmm…
#AI #VectorDatabase #RAG
Source
12:30 depending on what?
But how did you populate the vector database? Through another open-ai model vector model?
Thanks for the video mate
I wonder how useful this will be whilst replying to question specific or detailed doubts (such as while walking through the solution of a difficult math question a student might have a doubt about how did we go from one equation to the other). This would however obviously depend on the capabilities of ChatGPT and if it is not able to answer, what better could have been done?
Thank you Gaurav. Your Videos are really helpful.
When I hear gaurav, i hear system design 🎉🎉🎉🎉🎉
Hi Gaurav, Thanks for the video. I want to try with simple small book transcript in to vector db, augmented and use free model. Is there any free vector DB and LLM model which I can play with ??
Great video! I may have missed it, but did you talk about vector embedding and embedding models that turn this data into vectors to make a similarity search. Also wanted to know if neon automatically did that that.
Hey Gaurav, thanks for such a useful example. I am wondering if you do it again today, would you choose a different approach?
Solution being overplayed that they can replace Human Teaching assistant, if being realistic not possible in any way especially for Non-IT subjects.
Bro, how do you added
Data to vector db?
One question : How do we decide how many dimensions or what dimensions we need to have for our vector database?
Can anyone here answer this?
Gourav since very long i am waiting 4 ur free vedios…
The files upload is done every time a query has to be made? or ChatGPT retains that information against your account for future.
can u make a video on graphrag?
Hi Gaurav, excellent video. Thanks.. Quick question. How to query the vector database with the client's question to extract relevant transcripts?
Depending upon the salary😂
This will be costly with AWS. What a low cost option?
Hey Gaurav, good video. Not as good as your newer ones though.
07:00 minutes into your video and you have not mentioned RAG even once, not said why you are using a vector, not mentioned vectorization process. This was way too focused on vector DB of your experience w/o enough context.
Fan of your videos, btw.
Interesting. Thanks for sharing
Depending on your salary you can choose the Instance you want 🤣 What a guy !
Any reason why you did not use openAI assistant API that comes with inbuilt vector database?
As always, great content. Keep rocking.