Ramansh Sharma

MLE @ Omdena
💂‍♂️
Hi everyone! I'm a self taught Machine Learning engineer, looking to make the next big break in the theory of Deep Learning. Badminton and pizza are two of my favorite things, after Harry Potter of course :)

Live group coding session - Time Machine ⏰

💂‍♂️
Hey hey hey friends and community! Happy Friday!

Just sharing a few details about the upcoming group coding session happening on May 14th, 10:30 am EST for a duration of (more or less) 90 minutes!

Have you ever wanted to quickly lookup your meeting time often given in UTC into your own time zone 🤔❓ Googling that time is fast but no fun! Let's come together to build a much more fun and interesting solution for this 😃 and learn form each other at the same time.

We will build an interactive web app to recognize a drawn time into another time zone.

At the beginning of the event, we will split ourselves into subgroups to focus on different parts of the project -

  • Data processing - Taking the training image dataset (similar to any inference time test image) and pre-processing to feed it into a Machine Learning model.
  • Modeling / training - Training a Computer Vision model to recognize a drawn time into numerical digits.
  • Post processing - Converting the doodled time zone into another time zone.
  • Deployment - We want to try deploying this during the event itself if possible.

You can work on any part you are interested in! We will have engineers from the Deepnote team joining as well to help out in all of the parts! 😀

Rough agenda of the session:
[5-10 minutes]: The coders join in and move to the subgroup of their part of the project
[5-10 minutes]: Individual subgroup discussion on how to implement their code using different tools and methods
[55-75 minutes]: Happy coding folks 👨‍💻👩‍💻! Keep the discussion on with your teammates as you collaborate in the same Deepnote
[10-15 minutes]: Regrouping and putting all the different parts together to demonstrate how the Time Machine would work! 😃🎉

Some questions/suggestions that you might want to think about as we approach the date of the event -
  1. How will we take a drawn input from the user? - We are thinking of using Streamlit for taking a drawn input. Check out some of these resources to learn how to use Streamlit inside Deepnote
    Streamlit inside Deepnote
    Streamlit Canvas

  2. How will we train a Deep Learning model to recognize drawn time? - We can use the MNIST dataset for example to train a model to recognize handwritten digits. ⭐Find more tutorials to learn about Convolutional Neural Networks at the end of this post⭐
    https://www.kaggle.com/c/digit-recognizer/data
    https://keras.io/api/datasets/mnist/

  3. How do we convert time zones in Python? - There's this very useful library in Python pytz. This will come in useful to convert time zones in the post-processing part.
    PyTz

  4. Deployment? - Once we complete all the individual parts of the project, how do we put them together for inference locally, and then (hopefully) host it over Heroku for it to be live on the internet!
    Towardsdatascience tutorial
    Analytics Vidhya tutorial

This will be a community driven project, feel free to brainstorm more ideas here on alternative tools to use or even more questions to answer! Anything that can help us be more prepared for the live coding session 😁 The objective of the live coding session is to learn from each other in a friendly workspace and to make a cool project together! And of course, to have fun while doing so! 😄

Last but not least, here are some tutorials to get started with Computer Vision modeling with PyTorch and TensorFlow, two popular Python frameworks for Deep Learning. If you have any questions at all for the live coding session, let's discuss more in the comments or in the dm's.

Have a great weekend ahead folks!

TensorFlow
TensorFlow+PyTorch
To learn more about Computer Vision fundamentals, check out this great course form UC Berkeley
Like Comment

Cupy installation problem

💂‍♂️
Hi guys,

In my project I saw that cupy was not recognized so I tried doing 
!pip install cupy (as shown in the official pypi page)
but I'm getting a huge error message when I do that. Can I get some help on this if possible please?

My Resume

💂‍♂️
Hi guys,

I saw this channel for sharing résumés so I wanted to share mine! 😁

I have yet to add 2 of my newer projects in this résumé but they are live on my LinkedIn (https://www.linkedin.com/in/ramanshsharma/).

Would love to get any feedback at all regarding this! 🙂

Like Comment

Deepnote swag!!😃

💂‍♂️
Hi community,

I didn't know where else to share this 😂

Thank you so much  Anna Skoumalova  for shipping the swag to me in India! I absolutely love it so much (I wore the community t-shirt the same day I got the package 😁).

Like Comment

Publishing with Deepnote + Public profiles

💂‍♂️
Hi everyone!

I have made this short video on how to publish notebooks with Deepnote and introducing the public profiles feature. Do check it out 🙂

https://www.loom.com/share/b5cc60fcd0eb494da98e2e415353e237
Like Comment