Build and Host Real-world Machine Learning Services from Scratch


As Python has become the most popular programming language for research, it is quite easy to clone the source code from state of the art machine learning projects, and then run the demos in your well-prepared local environments.

However, is it easy to turn these demos into production-ready services? How can you make them sustainable? What actual challenges may you face?

In this talk, I will share a story about how we leverage the power of Python to build computer vision services for customers, and further improve these services through a self-designed machine learning pipeline.


Outline * Introduction * The first demo * Design a working prototype * Get alpha version on production * Kick off data collection pipeline * Make the service scalable * Build our machine learning pipeline * Develop more production-ready features * Recap and lesson learned In fact, this talk is a revised version (火力加強版) of my talk 'from ai.backend import python' in PyConTW 2018. I shared my story, what I have learned and my predictions about the future from a backend engineer's aspect. The talk was informative for backend engineers but not too interesting for other people, and so I've decided to tell the story across the timeline with some implementation detail instead of a still technical overview.




Joe (就是那個最近一直沒有好好辦台南拍的傢伙 QQ)