HN Explorer

Ask HN: Machine learning careers

13 points by gfr ↗ HN
I am interested in machine learning and natural language processing, but am slightly put off from devoting a significant amount of time towards it because I am under the impression that to make a career doing this, it is seemingly mandatory to have a PhD and you will be working in a research lab at a large corporation, the government, or a university. Is this true?

9 comments

[ 2.7 ms ] story [ 32.2 ms ] thread
carbocation
Instead of framing this from the negative, can you frame it in the affirmative. In other words, I know that you don't want to work in a large corp, the gov't, or a university. Where do you want to work?
gfr
I was a bit unclear, but what I mean is that can I live a life where I build things and get paid for it as opposed to spending time writing research papers and attending conferences, which I am under the impression requires getting a PhD to be taken seriously.

As far as where to work, anything from a startup to a large corporate company could potentially fit the bill.

carbocation
In that case, yes; I think there is plenty of opportunity to use machine learning (from relatively simple implementations to more advanced stuff) in many businesses. For example, I run a simple college admissions website, and we use some basic machine learning techniques. If we're using it, I imagine that nearly every field could take advantage of it in one way or another.
essrand
You can easily make a career in machine learning with a Master's degree, and a thesis that focuses on a ML problem.

To work as a scientist in a research lab, you do need a PhD, very few Master's candidates can make it there.

But to work in startups and some large companies doing Machine Learning mostly is not that hard.

I have worked in various companies as a "Research Engineer" and "Data Mining Engineer". I know someone who is a "Scientist" doing NLP in Yahoo Research and has only a Masters.

computerslol
Why do you have scientist in quotes here?
essrand
Just quoting job titles, since they usually don't mean much usually.
jbooth
If you're in NYC and know more about machine learning than just being "interested in" it, drop me a line at jbooth@proclivitysystems.com. We (along with several other startups) apply machine learning algorithms to big datasets for a business - most of us don't have PhDs.
monk_the_dog
I've worked in image processing and pattern recognition my whole career (~20 years). My last position was as principal software engineer. I don't have a graduate degree so it is certainly possible to work in the field without one.

Many places divide the work between two groups: an algorithms group and a software group. It's going to be harder to get into the algorithms group without a graduate degree (without experience anyway). But if you join a software group, there will be lots of opportunities to solve pattern recognition problems. And if your love is machine learning, you can move in that direction as you prove yourself to others.

I know nothing about natural language processing, but that's my $.02 from my pov.

PS. Hey jbooth: I took a look at your web site. You guys do cools stuff!

waterside81
As many here have answered, you don't need a graduate degree to get into this field and make money. Both myself and my partner at Repustate (http://www.repustate.com) have undergrads in CS, but other than that, we just read a lot of published papers from well known experts in the field, like Bo Pang @ Yahoo. She's a great source of information if you're interested in SVMs (support vector machines) and contextual polarity, that kind of stuff. Generally speaking, Cornell University has a really strong library of research in this area.