Sometimes when you upload Python binary packages to AWS Lambda from your local machine, you get an error. For example, if you want to use numpy with AWS Lambda you need a built version on AWS Linux.
Setting up AWS instance just for a build is a pain. That's why you can quickly install Docker and build any package on your machine. After building a package just copy into your vendored/ folder and upload to Lambda.
If you're ok with Python 3, simply build Docker image from a given Dockerfile with a command
docker build -t your_name/your_tag .
Don't forget to replace your_name and your_tag to whatever you want.
Then run your container and mount local folder with a desired requirements.txt file to a containers /workdir_container with a command
docker run --rm -v your_folder_with_requirements_file:/workdir_container your_name/your_tag
Now wait for packages to be built and find them at your_folder_with_requirements_file/vendored/.