When adding text to a raster-based image format like JPEG (as opposed to vector-based formats like SVG) then the text is no longer a separate object, it's just coloured pixels like the photo. Any transformations on raster images will almost certainly cause "blurriness".
To reduce the affects you could add the text last so it is not affected by the other operations.
I'm not sure what the GDI.net default level is but you could play with the level of JPEG compression. Low compression is not as lossy but produces large file sizes. You should be able to get away with 5-10% compression without losing too much detail while significantly reducing the file size, but that depends on the images and how fussy you need to be.