Docker efficiency of CI/CD pipeline - docker

Question

This is my travis.yml file and this is the latest run
https://travis-ci.com/github/harryyy27/allies-art-club:
sudo: required
language: generic
services:
- docker
stages:
- dev
- prod
jobs:
include:
- stage: dev
if: NOT(branch=master)
scripts:
- docker build -t harryyy27/allies_art_club/frontend -f ./client/Dockerfile.dev ./client
- docker build -t harryyy27/allies_art_club/backend -f ./src/Dockerfile.dev ./src
- docker run -e CI=true harryyy27/allies_art_club/frontend npm test
- docker run -e CI=true harryyy27/allies_art_club/backend npm test
- stage: prod
if: branch=master
before_deploy:
- docker build -t harryyy27/aac-client ./client
- docker tag harryyy27/aac-client registry.heroku.com/$HEROKU_APP/client
- docker build -t harryyy27/aac-src ./src
- docker tag harryyy27/aac-src registry.heroku.com/$HEROKU_APP/src
- docker build -t harryyy27/aac-nginx ./nginx
- docker tag harryyy27/aac-nginx registry.heroku.com/$HEROKU_APP/nginx
# Log in to docker CLI
- echo "$DOCKER_PASSWORD" | docker login -u "$DOCKER_ID" --password-stdin
- curl https://cli-assets.heroku.com/install.sh | sh
- echo "$HEROKU_API" | docker login -u "$HEROKU_USERNAME" --password-stdin registry.heroku.com
deploy:
skip_cleanup: true
provider: script
script:
docker ps -a;
docker push harryyy27/aac-client;
docker push registry.heroku.com/$HEROKU_APP/client;
docker push harryyy27/aac-src;
docker push registry.heroku.com/$HEROKU_APP/src;
docker push harryyy27/aac-nginx;
docker push registry.heroku.com/$HEROKU_APP/nginx;
heroku container:release client src nginx --app $HEROKU_APP;
I'm unsure as to whether this is an efficient setup. Is there a way of caching the docker builds so it doesn't have to rebuild them every time? When I'm updating builds locally it can update just the parts that need to be changed. Is there a way of doing this on Travis? A whole new rebuild seems like a lot of unnecessary work for Travis. What's the most efficient combination of free CI/CD service and free hosting for docker deployments in terms of build efficiency and deployment efficiency? (I know this is subjective, just want to get some idea)
Sorry if these are rookie questions btw I am very new to docker and its corresponding workflows/pipelines... having a bit of trouble finding multi container examples with heroku.