February 2024

February 7th 2024

Hello in February! I haven't really been posting very much, but I do hope to get in to a regular rhythm.

I recently discovered a new app called Obsidian it has been an interesting learning curve using it. Obsidian is a note taking app that gives you much more control over your notes, from accessible file storage to extensibility. I can save my notes to a google drive folder then access them any where. There's even extensions to let you execute javascript and query your notes to get data or display them so it's easier to parse them.

February 12th 2024

I think I fixed the link in the post above but it's a little hard to tell when I edit the files in Obsidian then post them up to S3 and build them on github. If it doesn't then I'll actually need to edit it in VS Code, I do need to work on this site more anyway.

February 13th 2024

I wasn't doing the above link correctly, hopefully it works now.

February 19th 2024

It does work!

February 21st 2024

I have been boldly attempting to use AWS SageMaker recently, in an effort to cut EC2 instance costs. I'm hoping that by applying autoscaling to my ML runs I'll be able to save a few bucks, and make it run faster.

February 22nd 2024

I stopped using SageMaker for the time being until I learn more about it, it sound's like there are a lot hidden costs attached to it.

February 26th 2024

I'm considering using AWS Textract to extract text from PDF files, like bills and such. It seems like a good idea, but I'm not sure I want to spend the money on it. Even though it would only cost $1.50 per 1000 pages of PDF. It seems like it could be useful for indexing documents I need to keep track of. I could even use it for JPEG images for other documents as well. They also have some other tools like Rekoginition to tag my photos. I need to figure out if it's something I want to go through the process of building my own or using their already built tools.

February 28th 2024

I started a project to send documents to Textract and Rekognition, however they need to be in S3 already which caused me some confusion as I was working on it. Also PDFs have to be sent to Textract asynchronously, which means a waiting period that I will need to find a way to handle. Since this all happens in AWS I could potentially do it in a lambda that runs infrequently enough so that I won't get charged very much.