demo at https://www.youtube.com/watch?v=LTzNR9FVliU preinstallation of quicklisp is required, to run the project the files have to be placed in ~~/.quicklisp/local-projects/cldml/~:

[ ! -d ~/.quicklisp/local-projects/cldml ] && mkdir -p ~/.quicklisp/local-projects/cldml/
cp math-utils.lisp ~/.quicklisp/local-projects/cldml/
cp utils.lisp ~/.quicklisp/local-projects/cldml/
cp network.lisp ~/.quicklisp/local-projects/cldml/
cp regression.lisp ~/.quicklisp/local-projects/cldml/
cp cldml.asd ~/.quicklisp/local-projects/cldml/

now the library cldml can be imported in the repl with (asdf:load-system :cldml) to run the project:

file=server.lisp
sbcl --eval '(push *features* :hunchentoot-no-ssl)'\
     --eval '(ql:quickload :cldml)'\
     --eval "(load \"$file\")

then go to http://127.0.0.1:3000/client.html

the code is documented at https://mahmoodsh36.github.io/convolutional_neural_network.html