Here's What's Cooking At Segmind - Introducing Segmind Recipes

Machine Learning Apr 20, 2022

There is an explosion of new algorithms being developed in domains popular in AI like computer vision and NLP. The innovations and fast-paced development is mind boggling. And for an ML scientist/engineer, playing around with new exciting algorithms and getting immersed in projects while exploring and digging deeper is the dream. But today, there exists a lot of friction and time lag between learning about these algorithms and the implementation of them. You have to first search for the right code, then clone it to your local/cloud environment, setup the hardware and environment according to the needs of that algorithm, make sure you have the right compute requirements and then start replicating the code. The above meme depicts the way ML folk feel after going through all this exhausting process. Wouldn't it be easier if you can start working with these new algorithms with just one click of a button?

Introducing Segmind Recipes - curated cutting edge new algorithms that you can start working on without the worry of setting up the code, environment and compute. Pick any algorithm and start working on it, in just one click.

We are inviting data scientists, ML/DL engineers/scientists to add Recipes to our platform. When you add Recipes to our platform, you'll be rewarded with ☁️ credits and some Segmind goodies/swag 😎

Below are the steps and criteria to add Recipes:

Picking A Recipe ✍️

  1. Go to https://tinyurl.com/segmind-recipes
  2. Browse through the existing Recipes to make sure you pick one that has not been added already.
  3. Go to this Google form and submit your Recipe idea https://forms.gle/NKD9EBEyBebRH4WA7

Creating The Recipe 🥣

  1. Sign-up/Login into cloud.segmind.com
  2. Create a new project by clicking on the + button on top right
  3. Next, create a new instance as shown below.

Submitting The Recipe 👀

After you finish the Recipe and make sure everything is working well, share the instance with harish[at]segmind.com. We will run the Recipe on the selected environment internally and evaluate that the code is running as expected. If everything works well, we will add it to our Recipe directory.

Criteria 🧐

  1. The algorithms should fall broadly under these domains; Computer Vision, NLP, Reinforcement Learning, Speech/Audio and Time Series.
  2. From an industry standpoint, they should fall broadly into the following categories; Medical, Autonomous Vehicles, Fintech, Retail/Consumer and Remote Sensing/GIS.
  3. From a framework standpoint, they should either be in TensorFlow, PyTorch, XGBoost or SpaCy.
  4. Dataset size should not exceed 20GB (if more storage is required, please reach out to us).
  5. The algorithms should be able to run on single NVIDIA T4 instance.
  6. It should not take more than 10-20hrs of NVIDIA T4 GPU compute time to run the recipe.
  7. All the pip and apt get packages must be installed through the Jupyter notebook interface not Terminal.

Rewards 🎁

Upon signing up for adding Recipes, you will get $10 in cloud credits to develop the recipe. Once we evaluate and add the Recipe to our directory, you'll be awarded $5/Recipe. If you add 5 or more recipes, we will send you some Segmind goodies. Share your Recipe on any social media platform (Linkedin, Twitter, Reddit), we'll add $5 more in cloud credits. The more recipes you add, the more cloud credits you earn.

Tags

Harish Prabhala

Harish is the co-founder and head of product at Segmind.