For the last few years, I have quietly been developing a tool for self study using M.I.T's Opencourseware as the primary resource.

Now I'm applying some of techniques I learned in #ProductManagement and #SoftwareEngineering to the fantastic roadmaps.sh

You might also think of this as an exercise in #LiterateProgramming.

Create a set of bookmarks for further reading and some documents to ingest into our machine learning model.

for n in $(seq 100 123); do
    find src -name '*.md' -type f -path '*backend*' -and -path "*content/$n*" | xargs awk 'FNR==1{print ""}1' > $ROADMAP-$n.out
    pandoc --standalone $ROADMAP-$n.out | pup 'h1, a json{}' > $ROADMAP-$n.json
    pandoc --standalone $ROADMAP-$n.out | pup 'p text{}' > $ROADMAP-$n.md
    rm $ROADMAP-$n.out
done

Each of the URLs in the bookmarks match a target in the Makefile. This will hydrate/inflate the amount of text we are creating.

for target in $(cat $ROADMAP-100.json | jq -r .[].href); do make -n $target; done

Now ingest the content

python3 ingest.py

Bookmarks tagged h1 are good candiates for questions

cat $ROADMAP-100.json | jq '.[] | select(.tag == "h1").id'

The first thing you need to do is install zx and run zx README.md. That will execute the code in this file and advise you on the next steps. We're using terraform and Google cloud platform to spin up a server capable of running the models

terraform apply

You should be able to ssh into the instance.

GPG_CONFIG ssh

It's clearner to wherever possible, spin up and down on an as needed basis. This is a more precise style of development. Can't hack what don't exist.

terraform destroy

I've created a project with chromatic to get feedback on the visual changes. GitHub codereviews are a good place to handle everything else.

Follow along for more or request access at linkedin page