@Medampudi@awscloud@credly Thank you very much π.
You are correct. I think companies use reserved instances and saving plans to approximate thier future compute cost, forgoing the flexibility and accepting some capasity waste.
Just passed my @awscloud AWS Solutions Architect Associate certification π
The depth of AWS services to actually understand (not just memorize) was the real challenge β VPC design, IAM intricacies, cost optimization tradeoffs.
https://t.co/txKmdPaS43 via @credly
Just passed my @awscloud AWS Solutions Architect Associate certification π
The depth of AWS services to actually understand (not just memorize) was the real challenge β VPC design, IAM intricacies, cost optimization tradeoffs.
https://t.co/txKmdPaS43 via @credly
@ROHITMH_1@Om7248 If you want to just quickly test, or have someone test, you can use your computer with say ngrok.
If you want to running somewhere else you can use Render or CloudRun or .....
There are different options. It's more about your requirements and opportunity cost.
@ROHITMH_1@Om7248 You method of deployment is project dependent if you're trying to reduce cost.
For example, if the project is not too big and doesn't run for a long time you can use something like AWS Lambda (make sure to setup budgets so as to not pass the fee tier).
Batch the diff from the Pull Request when the PR is long, else don't.
Approximated token value per char and calculates the pr length by tokens then set the maximum token per batch.
Working on the Ai Code Reviewer:
Thinking of how to best balance API limits and review properly.
I can either:
- Review each change per file for each file = fragmented review + LLM API Rate Limit OR
- Review the whole change at once - LLM token limit
My decision: