YT Comment Sentiment - Backend Side

Technology	Description
	A Python library for building and training machine learning models.
	A collaboration platform for machine learning, hosting data and MLflow models.
	A platform to manage the ML lifecycle, including model tracking and deployment.
	A modern web framework for building APIs with Python, known for its speed.
	A Python library for data validation. Used to validate API data.
	A testing framework for Python, used to test the FastAPI application.
	An API to access and manage YouTube video data, including comments.
	A cloud platform for hosting APIs, websites, and applications.

What I Followed to Know?

Important

As I am learning Python, Data Science and Machine Learning for more than 3 years. I don’t have to look around to learn new things to build this. This part is kind of easy for me.
But as I said earlier, the documentations and ChatGPT is most important resources you can onto. :wink:

Need to get an API key from Google Developer Console to interact with YouTube Data API.
Need to create an account on DagsHub to store/track MLFlow experiments and models.
Created a DVC pipeline to run the MLFlow experiments seemlessly using dvc repro command.
After creating the FastAPI app, I’ve used pytest to test it and also setup a pre-commit for it.
Deployment on render.com.

What Type of Problems I Have Faced?

Render.com

As I have used uv to manage my project but render.com doesn’t support uv out-of-the-box so I have used pip to use uv for dependencies installation.

pip install uv && uv sync --extra=backend --compile-bytecode

Also, render.com only serve apps on port under $PORT env (which 10000 most of the times) so make sure to explicitly provide while running app through uvicorn or fastapi-cli CLI.

# For uvicorn
uvicorn run --host 0.0.0.0 --port $PORT backend.app:app
 
# For fastapi-cli
fastapi run --host 0.0.0.0 --port $PORT backend/app.py

Docker

I am using wordcloud to create a plot in a FastAPI route. While building docker image FROM python:3.11-slim image, I am getting error because wordcloud package needs gcc package to build wheels. So you need to explicitly install gcc before install wordcloud as python package.

# Install gcc for wordcloud
RUN apt-get update && apt-get install -y gcc && apt-get clean
 
# Now install project dependencies including wordcloud
# ...

Also use multistage builds in Dockerfile to reduce the image size. See uv docs.