Showcase your ML models like a pro

3 min readAug 26, 2022

Well someone said correctly that ML models in local are just storage eaters. After months of training, preprocessing and whatnot and still that model remains in the local machine, it's not justice. In the current era, there are a lot of advancements going on in ML deployment platforms and one such platform I came across recently is DAISI. It is the foremost destination for developers like us to put their models into life and expose them to the world so that everyone/anyone can benefit from it.

Introduction

I recently developed an Abuse detection model in Indian languages to catch offenders on platforms like Twitter, Facebook, and other social media apps. With growing Indian internet users across different social media apps, ML models like this can be put to great use.

In vanilla terms, my model takes input as text and language and returns the probability if that text is abusive or not.

Daisi platform

Deploying model on daisi

Bundle your code into the python functions and these functions will act as an endpoint. You can also make a streamlit app for UI which ll be shown on daisi dashboard. Make sure you have mentioned all the necessary packages that are required to run the app in requirements.txt. Push your code to the GitHub

2. Click on “Create a New Daisi”, at the bottom right of the Daisi platform UI

3. Enter the path to the GitHub repository, Select the branch, folder and main Python script. Click on “Create”

Wooaha!!! Daisi is this simple to deploy. It will look like -

Inferencing

Daisies can be called from anywhere. In Python, the pydaisi the package helps Daisies calls as straightforward as if they were local functions, passing arbitrary Python objects in argument and receiving whatever the function returns.