I found myself:
- Needing to find a new job
- Wanting to use Kedro more
- (Probably) needing to add some stuff to my GH profile
So I made this.
A Data Engineering-heavy collection and processing pipeline gather's HN Who Is Hiring posts, with minimal analysis.
Posts are gathered, by one or more months, and semi-filtered for provided keywords, such as data scientist, machine learning engineer. No fuzzy matching is done.
This repo runs on kedro. Requirements are standard python (~=3.10) and an appropriate environment.
- Clone the repo:
$ git clone https://github.com/BrianCechmanek/hn_hiring.git - In terminal of choice change to project dir:
$ cd hn_hiring - (optional, recommended) create a python 3.10.5 venv via pyenv:
$ pyenv virtualenv 3.10.5 hn - (optional, recommended) activate venv:
$ pyenv activate hn - install kedro & requirements:
$ python -m pip install -r src/requirements.txt - run pipelines:
$ kedro run - (optional, highly recommended) visualise the ppl:
$ kedro viz(opens browser window. refresh after a moment if you get an error page)
Outputs will be in data/*
- (optional, recommended) run streamlit viewer:
$ streamlit run src/app/viz.py(opens browser window)
The code is strictly split into discrite pipelines. Each can be run separately, though an initial sequential run is required (to collect and process the data).
The Kedro data convention is similar to that of cookie-cutter data science's, where raw, primary, ... , reports, are explicitly separated from each other. I follow this.
The full pipeline is concatenated such that kedro run runs all three sequentially.
Data are collected via the public HN API.
- To run:
kedro run --pipeline=data_collection - I suggest only running this pipeline once or twice. it takes a minute, but mostly I don't think it's nice to pummel the HN API (though, really, only a couple hundred calls are performed)
- See pipeline specifics in it's README
- To run:
kedro run --pipeline=data_processing - See pipeline specifics in it's README
- See pipeline specifics in it's README
- To run:
kedro run --pipeline=models
At this time, I'm not training, running, or evaluating any models. I May add something in the future.
I do not use notebooks at this time.
A very minimal streamlit app is writen to scan through the resulting posts. It covered my bases, so I stopped there.
- I don't think the "freelancer" or "wants to be hired" posts have any relevance
- pre-commit is used for formatting and linting. Feel free to install it, if you're going to do anything to the code. I just like using it
- If you wish to run tests, you can run
$ pytest .. You will have to install the package:python -m pip install -e .followed bypytest .(or whichever selection you desire) - I wanted to do some async stuff with the API collection, but kedro doesn't handle it natively, and writing an AsyncRunner was WAY out of scope
- I read somewhere(!) that the way to go about HN posts is through big query, but as of Feb 2024, my weak googling only pulls up results for old (2015) datasets there. If I can find, I'll re-write the
data_collectionpipe for BQ instead - Children of posts are ignored: in general Who's Hiring posts shouldn't get replies; their value would be unclear; I wan't to avoid a recursion task at the moment
- Sometimes a poster will reply to themselves with additonal job info. Looking at you Lloyd's. Thus, I'm not worried at this time about losing descendants.
(the below is the stock Kedro README. I've kept it for reference)
This is your new Kedro project, which was generated using kedro 0.18.12.
Take a look at the Kedro documentation to get started.
In order to get the best out of the template:
- Don't remove any lines from the
.gitignorefile we provide - Make sure your results can be reproduced by following a data engineering convention
- Don't commit data to your repository
- Don't commit any credentials or your local configuration to your repository. Keep all your credentials and local configuration in
conf/local/
Declare any dependencies in src/requirements.txt for pip installation and src/environment.yml for conda installation.
To install them, run:
pip install -r src/requirements.txt
You can run your Kedro project with:
kedro run
Have a look at the file src/tests/test_run.py for instructions on how to write your tests. You can run your tests as follows:
kedro test
To configure the coverage threshold, go to the .coveragerc file.
To generate or update the dependency requirements for your project:
kedro build-reqs
This will pip-compile the contents of src/requirements.txt into a new file src/requirements.lock. You can see the output of the resolution by opening src/requirements.lock.
After this, if you'd like to update your project requirements, please update src/requirements.txt and re-run kedro build-reqs.
Further information about project dependencies
Note: Using
kedro jupyterorkedro ipythonto run your notebook provides these variables in scope:context,catalog, andstartup_error.Jupyter, JupyterLab, and IPython are already included in the project requirements by default, so once you have run
pip install -r src/requirements.txtyou will not need to take any extra steps before you use them.
To use Jupyter notebooks in your Kedro project, you need to install Jupyter:
pip install jupyter
After installing Jupyter, you can start a local notebook server:
kedro jupyter notebook
To use JupyterLab, you need to install it:
pip install jupyterlab
You can also start JupyterLab:
kedro jupyter lab
And if you want to run an IPython session:
kedro ipython
You can move notebook code over into a Kedro project structure using a mixture of cell tagging and Kedro CLI commands.
By adding the node tag to a cell and running the command below, the cell's source code will be copied over to a Python file within src/<package_name>/nodes/:
kedro jupyter convert <filepath_to_my_notebook>
Note: The name of the Python file matches the name of the original notebook.
Alternatively, you may want to transform all your notebooks in one go. Run the following command to convert all notebook files found in the project root directory and under any of its sub-folders:
kedro jupyter convert --all
To automatically strip out all output cell contents before committing to git, you can run kedro activate-nbstripout. This will add a hook in .git/config which will run nbstripout before anything is committed to git.
Note: Your output cells will be retained locally.
Further information about building project documentation and packaging your project