LinkedIn²

Search smarter not harder.

LinkedIn² is a utility to rank jobs according to desired keywords or features, for example:

(javascript^2 OR data scien*) AND NOT ("java" OR "fluent german"~2 OR manage* OR ([5 TO *] years))

Also keep track of already seen jobs, already applied ones and (really) new posted jobs.

It's true that LinkedIn already allows boolean search, however, it doesn't have all the power of elasticsearch's query string.

And not only that, it also allows taking features like company's employee count and job posted date, then rank higher those posts that meet a certain threshold for them.

First, install my own fork of the linkedin-api package

pip install git+https://github.com/4erf/linkedin-api.git

It tries to circumvent the 1000 job limit of LinkedIn's voyager api.

Then, on jobs folder, create a .secrets.json which contains your LinkedIn accounts, the more accounts the faster the fetch process and less likely you get rate limited.

[
  ["email1@domain.tld", "pass1"],
  ["email2@domain.tld", "pass2"]
]

Now you can modify the include and exclude arrays in get_jobs.py, together with job_type, location_name, and other parameters received by the get_all_jobs function.

This allows you to run a pre-filtering using LinkedIn built-in search, to facilitate the fetch process excluding definitely irrelevant jobs.

Now, on the root folder, run

docker compose up

to run both elasticsearch and Kibana nodes.

You can then log into Kibana by going to localhost:5601 and activating manually using elasticsearch's address http://es01:9200. This is just in case you want to have Kibana access, not needed to use the app.

Now to fetch all jobs matching the query, run:

python get_jobs.py

This will first try to get the highest % of jobs possible, circumventing LinkedIn limitations, when you reach a reasonable %, press Ctrl+C and now wait for the job details to be fetched.

After finished, it will give create three json files, jobs_src which contains just the initial list of jobs, jobs which contains jobs with all the additional information added to them, and jobs_nd which is used to insert the data into elasticsearch.

These files are just in case you want them for other purposes, the jobs will be on elasticsearch already, ready to use.

Now, to visualize jobs and interact with the system in general, run

yarn install

on the front folder, and then run

yarn start

This will fire up the web app, and you should be able to see your list of jobs, apply filters, mark as seen, applied, etc.

• Automatically fetch new jobs, right now the process has to be repeated, though it's possible to only fetch jobs posted after certain date, reducing the amount of API calls needed.
• Simplify installation process, automate as much as possible, maybe containerize everything