adityajn105 / udacity-mlops

Udacity MLOPs Nanodegree Project

ML pipeline to expose API on Render

Udacity MLOPS nanodegree project to train a model on census data and publish it via API on Render

Github Link for Project: https://github.com/adityajn105/udacity-mlops/

Developer Environment

1. Have python3.8 installed.
2. Firstly install all requirements: pip install -r requirements.txt
3. Build models using python mlpipeline.py
4. You can test using pytest

EDA

EDA notebook is provided: link. This is referred while developing cleaning procedures.

Running individual Steps:

Cleaning Steps

To clean the data, it will create cleaned_census.csv in data folder. python mlpipeline.py --step=clean

Train/save model

To train model, it will create necessary model files in model folder. python mlpipeline.py --step=train_save_model

Slicing evaluation

Evaluate model on different slices, output can be found in data folder. python mlpipeline.py --step=evaluate

Run all steps one after other

python mlpipeline.py

Check if Render server working

python check_render_api.py

CI/CD

This project is supported by CI/CD, every new commit to main or pull request to main will trigger a github action workflow.

Link to CI pipeline

For Rubriks

1. live_get.png
2. live_post.png
3. example.png
4. Model card
5. continuous_integration.png
6. continuous_deloyment.png