🤗 Huggingface Spaces: Ideal place to show your skills as newbie Data Scientist/ML Engineer

Showcase you AI skills by hosting your models at Huggingface for free

🤗 Huggingface Spaces: Ideal place to show your skills as newbie Data Scientist/ML Engineer
image by dalle e (by author)

There are times when everyone wishes they could showcase their Data Science skills by hosting their model somewhere for free, allowing interviewers, stakeholders, or anyone interested to test the model in real time with their own dataset. Using a GitHub repository is one way to do this, but there are some issues. For instance, sometimes even installing the dependencies on your personal computer can be a real headache. Additionally, it takes a lot of time, and if it does not work, it feels like a waste of time.

I found myself in a similar situation when I was in Paris this year to attend the International Conference on Computer Vision (ICCV). My excitement doubled upon learning that Hugging Face was organizing an Open Source Meet at Station F. This event was particularly special because they invited everyone to demonstrate their work. I applied there, submitting the name of my poster (which I was also presenting at ICCV, thinking it would be a great opportunity to meet and socialize with others). However, I was slightly concerned about how to demonstrate that my model works on a variety of images; as a student, I didn’t have substantial funds to pay cloud companies for hosting my model. After some research, I discovered Hugging Face Spaces and instantly became a fan. Here, you can host your models for free (albeit only on CPU), and there’s a Python library called ‘Gradio’ that allows you to create an impressive UI for your model (as shown in Figure 1). By the way my demo was a huge success at the event, and I made tons of new, like-minded friends there.

Figure 1. An example of UI hosted at Hugging face (image by author)

One more thing: you have two options for showcasing your work. You can either present the demo on the Hugging Face website (as illustrated in Figure 1) or display it elegantly on your personal website or GitHub page, just as I did (as shown in Figure 2).

Figure 2: Presenting demo at Huggingface event at station f

Additionally, everyone who presented their demo work at the event was featured by Hugging Face on X (we are at 5, YOLO-SAM). Please take a look at the tweet below.

This was my first time at ICCV, and it was memorable as I had the opportunity to present my model both orally and through practical demonstrations. Since I stammer, demonstrating something experimentally is more enjoyable for me than just talking. Now, I encourage you to go ahead and build something cool to showcase. Here are the relevant links:

Hugging face space: https://huggingface.co/spaces
Gardio: https://www.gradio.app/
Embedding on webpage: https://huggingface.co/docs/hub/spaces-embed

Happy building 😊