300 points by data_scientist_anon 6 months ago flag hide 15 comments
hacker1 6 months ago next
Impressive work! I've been looking for a well-documented real-time machine learning API.
hacker2 6 months ago next
How did you handle data preprocessing? Did you use any existing libraries for that?
hacker3 6 months ago prev next
What framework did you use for machine learning? I'm curious because I'm a TensorFlow user.
hacker4 6 months ago next
Yeah, which ML library or framework did you use? I use Scikit-learn for my projects.
hacker5 6 months ago prev next
I chose PyTorch for this project. It suits my use case much better than other frameworks.
original_poster 6 months ago prev next
Thank you! To preprocess the data, I created my own set of functions and made use of the Pandas library.
hacker6 6 months ago prev next
How long did it take you to build this from scratch including testing and documentation?
original_poster 6 months ago next
It took me roughly around 6 weeks to get from concept to a final, production-ready version.
hacker7 6 months ago prev next
What were the toughest challenges you faced during the ML API development process?
original_poster 6 months ago next
The hardest challenge was ensuring data consistency when conducting real-time preprocessing. And handling the automatic scaling.
hacker8 6 months ago prev next
How can we connect to the API and test it ourselves? I couldn't find a documentation link.
original_poster 6 months ago next
You can find the documentation here <https://github.com/your-username/repo-name>. Thank you for the question!
hacker9 6 months ago prev next
Are you open-sourcing it? I'd like to contribute or learn from the source code!
original_poster 6 months ago next
Yes, I'm happy to say that it is open source and available on GitHub under the GPLv3.
hacker10 6 months ago prev next
Excellent work! I'm glad that you shared your experience and code with the HN community.