GenAI Comics, Coffee, and Podcast (1/3)

I participated in the Kaggle: 5-Day Gen AI Intensive Course with Google.

It was fun juggling between work and these fun segments.

I wanted to create something cool, but as time has always accompanied everyone, we can only make as much as we have. So, let’s dive into creating a podcast and comic from a code file.

The first challenge was: what to make? I then thought of a problem: what if your org is not supportive of tools like notebookllm but supports LLMs? So, can we do something for ourselves?

The answer is, thankfully, yes!!!

So, how do we do that now?

I had been wondering how Cursor and Windsurf take care of our code. It turns out that, after a bit of research, those guys have an extensive mapping of the codebase and the functions within it.

So, how do we replicate that? Thanks to open source, we now have libraries to go to for those.

Although I have used NetworkX here as available in Python, Cursor uses a bit more sophisticated tools.

You can refer here:

ChatGpt search, I had too much clutter in my Gemini deep research, sorry! I will strongly suggest using that, though

So, how do we do this next?

Here, we would code and come up with implementations for these… Nah, just review the functions by Gemini and see how that helps save time. ( Now, is this project AI-made or human-made? Its secret )

Next, we had this good starter Kaggle book for calling the Gemini API from Kaggle:

Kaggle Model

Once we press New Notebook, we are ready to go and start collating things.

But we are only starting up. In the next blog, we will address the challenges of using TTS in Kaggle, getting the output to a podcast, and finally generating a comic.

Reference Kaggle book:

https://www.kaggle.com/code/vasusoni/kagglecodepod