Ask HN: What's your learning path for ML/AI developer?

2 points by aswathrao ↗ HN
How long it you took you to understand to work in ML/AI technologies (to get a job or compete in kaggle) and what learning path(books/courses) you finished ?

and any advice you want to add !!!!!!

3 comments

[ 3.6 ms ] story [ 14.4 ms ] thread
For me it was random.

I got a PhD in physics, where the job market was brutal. (I don't know if I was in the top 3% in talent but I certainly was not in the top 3% at either office politics or musical chairs.)

I got into software, did Perl, PHP, Java, ColdFusion, whatever paid. I started developing advertising-supported web sites that gathered content through an "A.I." pipeline just before Deep Learning broke in the news. That got me work on text analysis, productizing a deep-learning based search engine and some other work. I also wound up chasing some dead ends and going sideways at times -- at the moment I am doing pretty much ordinary front end but I am focused on getting really good at it.

(Something I learned the hard way is that real A.I. systems are going to have to work closely with humans. If somebody wants you to build an A.I. that does a 'difficult' or 'impossible' task the first thing to do is build a U.I. that lets a person do the task manually, otherwise where are you going to get the training set from? Starting with anything else is outright malpractice.)

I've been watching some videos on fastai
FastAI -> everything I can find on PyTorch -> pandas/numpy deep-dive -> wrangle, build, rinse and repeat.