Coding an MLOps 🦄

Week 15

I know, week 15? This is the first day I decided to do any blogging, so if you are the one person reading this, apologies for skipping so much. On the bright side, there should be 6 more of these weekly postings before going live.

Why write a blog? I decided to do this because mainly, I want to keep a log of all the things going through my head and sometimes need a break from code, reading about lean canvases, and cap tables. And what is with the unicorn 🦄? Well, if nothing else, it makes me laugh and makes for a nice emoji 😬.

For this post, I thought I’d tell the backstory why this is a problem I thought was worth solving and a warning to the nontechnical folks that none of this will make any sense. In 2018 when learning ML via Fastai, one of the lessons on deployment felt clunky, not because of the course content, but because there didn’t seem to be any reasonable solutions for easy deployment and inference via the web. Fast forward a couple of years, in 2020, lesson 3 of Fastai had pretty much the same problem. And this wasn’t unique to Fastai, other courses are the same. So I thought I’d give it a shot.

Thanks for reading and another update will come next week, focusing on some of the current technical challenges.

By the way, I will be looking for feedback. Please email me at info@mclabs.me if you are knowledgeable in ML and want to try it out.

FastAI student, MC Labs founder