My favorite comments from @sherwang, on a recent call as we walked through what's needed to deploy a model on the @DescartesLabs (DL) platform (see thread ...)
- "This is so much easier than having to download thousands of tifs in order to deploy my model!"
- "The platform exports imagery directly to ndarrays, so I can just run my code in python?" (yes!)
- "The platform exports imagery directly to ndarrays, so I can just run my code in python?" (yes!)
- "I can just load in trained model weights and deploy the prediction directly?" (yes!)
- "Wait I don't even have to break up my region of interest into smaller tiles myself?!" (you don't, tile definition is easily automated)
- "I am so thankful this company exists!" (me too!)
- "Wait I don't even have to break up my region of interest into smaller tiles myself?!" (you don't, tile definition is easily automated)
- "I am so thankful this company exists!" (me too!)
...My favorite reactions to the DL platform come from people like Sherrie that know the *true pain* of having to do the data wrangling and implementing of large deployments themselves. I'm ever so excited to work at a place that enables scientists to get to their science quicker.