Agree with the exhortation to just go do something (or just get a job doing what you want to).

Studying doesn't give you the same experience as someone who's been working in the field for 10 years *or* the experience of someone who's done a PhD.

If you want to compete with PhDs, you need to write papers. If you want to compete with people with experience, do something.

To the extent that there's a widespread/strong bias against autodidacts it's probably because there's a misalignment in what is interesting to learn and what is useful to an employer.

I'm a CS student getting ready to start applying for my first co-op/internship and could use some guidance from you more experienced folks..

My resume is pretty lacking (an unrelated BA degree and many years of poker which I'll probably not mention) so I'm trying to fill it out with lots of small technical projects. While taking my uni courses, I've been following Udacity's Fullstack Developer NanoDegree, which is comprised of several courses on different aspects of webdev. Each course finishes with some kind of project; many of them are fairly small, and sometimes we're building on some basic code that was already provided to us. So my question is this: is it okay to still list these projects on my resume? I'm certainly learning a lot from them, and have spent a fair bit of time putting some of them together, but don't want to create the impression that I've built them from scratch. Also, how does one present non-production apps and projects to employers? Is it enough to just say "here's my github, look at my code" and to be able to explain it in an interview, or should I only be listing projects that I can actually demonstrate live?

you should have 1 simple working app that's easy to play around on, just host it on heroku if you need a server or github pages if it's pure frontend

we got about 500 resumes for 1 internship slot, i don't think we looked at anyone's github

I also think there is a discord between perceptions and reality, in part.

There are basically two classes of auto:

person 1 = the Codecademy / Udemy / W3Schools crowd.

person 2 = self full-on education (or as one auto said to me: a self-taught polymath).

The fact is, most people who call themselves autodidacts are going to be in slot one. While many of them were in college, etc, it is a very different creature than person 2.

The problem with person 2 is that (I think leavesofliberty is this type), he or she honestly believes it is a great story, showing initiative, self-starting, passion, etc. The reality is that hiring managers are thinking "cool story, bro."

Because there is a much larger dominance of person 1, and person 1 has did just enough to get in after 6 months of learning, a hiring manager is going to have the perception that all autodidacts are person 1. Since person 1 has already burned the trail, it stands to reason that person 2 should set aside his ego / self-doubt and follow the path of person 1. Person 1 is at an advantage because they know they are impostors (of sorts) and embrace that fact.

The problem with person 2 is he is frightened of being an impostor. He goes out and builds up a website with user logins, commenting, rare frameworks, plug-ins, loads up ads, creates a massive neural net, etc. This is all just one little step from monetization.

What does person 2 end up looking like? An overly driven person who plans to bide his time at a company, put in said 40+ hours at night, then quit once he figures out his strategy. This is a high-risk person, and while it sounds wrong from an idealistic perspective (and contradicts the 100s of HN front-page articles on "passion"), it is not wrong from the employer perspective. They don't have time to deal with people who aren't going to stick around and commit to the growth of the company for a few years.

Of course, there are other problems with the person 2 story. One that I've dealt with a few times is utter skepticism. Did I really teach myself everything I claim? Did I really build whatever by myself? That's a large ball of yarn to swallow, and it doesn't make a lot of sense to the interviewer or recruiter. They can quiz me all day on this stuff and I'll be sure to answer their questions, but the point is people understand simple stories, and it does well for person 2 (or anyone with initiative to roll their own) to understand this.

tfw you get to your hotel in SF that your wife booked and you realize its in the middle of the tenderloin

Happy Independence Day to Americans (and Republic Day to Filipinos)!

Disclaimer: I've never written an ML program yet

Just out of curiosity, why do you need specifically C++? Is that because you're dreaming about writing a chess engine (or an AI for some other game) where the speed of computation is critical?

Well, though it's not impossible to write a chess engine beating the current top 3 (Stockfish, Komodo and Houdini) by means of cleverly designed neural network(s), it would be rather an academic success than a lucrative business - there's too little money in chess and in skill gaming in general, and online poker might run out of money soon too.

A quick search has pointed me at a Quora discussion about the comparison of C++ with Python and R for ML purposes. While there are C++ ML libraries like mlpack and Shark, you need to check if they have the functions that you need. Rewriting standard routines on your own isn't fun.

Anyway, good luck with your dreams!

Easy crack purchase - no walking. She's looking out for you.

I never really thought about it from that perspective. For a person to say screw it im going to go and try to learn this and start anew is quite a steep hill to climb. Having said that, I think all the low hanging fruit has been picked for the early adopters who did the bootcamps a couple years back. Now like any market, it's starting to get more competitive even for the basic junior/entry level web dev/ software developer. I was just talking to a TA at a bootcamp and he was commenting to me that essentially all the junior level positions got taken up by all the previous graduates and now that demand was met fully the next coming graduates will have a tougher market (that's including me). Fair enough. I'm ready to learn and put the effort to gain more skills and knowledge.

Having said that I still hold that just because you graduated from the course that does not make you a "coder" "hacker" "software dev" etc. It makes you someone with just enough understanding of what you still have to learn and as long as you go out to learn and work at it you will reach that spot. I unfortunately think that for the people I'm still seeing graduate from these courses fully think they are ready to start getting paid $60k+ without much effort and very little understanding even of what a AJAX call is, traversing the DOM, or much less telling you how to empty an array into a different one.

Then again, I know one girl who got hired as QA tester for company down here at 55k+ benefits at a low cost of living area who for her final "project" essentially presented a static site with a collage of jpegs to even include a jpeg of a google maps API frame.

This sounds like BS. I would try to contact as many previous grads and ask how long it took them to get jobs. It sounds like he is letting you know that getting a job will be hard, but wants to frame it so they still look good.

No chance you're wondering around looking to meet attractive women. Smart women.

This. It is far more important to be able to talk about one project well (design, implementation, challenges, next steps, etc.) than to have it (or a bunch of little ones) on github. Most of your interviewers will spend somewhere between 0 and 2 minutes looking at your github before you talk.

Sure--why wouldn't they? In all of that text, it isn't clear to me what the advantages of said "polymath" is over someone who, you know, went the normal education route and then did something (either on their own or working for someone else).

Indeed it points to the opposite: someone who is likely to be motivated/value to "do it their way, on their own" which isn't typically what one wants in a colleague.

Someone telling me that they are a "self-taught polymath" is definitely in "cool story bro" territory and probably somewhat worse than that if they don't have anything to show for it.

For the self-taught people, what's your measuring stick of success? How do you know when you've learned something?

For people that go the school route this is work they do that gets judged by other people (assignments, projects, exams). And I guess, ultimately, the degree. The degree says "this institution thinks this person learned some minimum threshold of knowledge about X".

And when talking about advanced degrees, it also becomes things like published research, thesis defended, etc.


(Disclaimer: I'm not claiming the measuring sticks used by schools is particularly effective.)

Personally, I think designing, and building useful applications with the new knowledge (though this approach can be perverted into "hack what I can find"). Build something that shows orginality, organization, enginuity, and mastery.

I am going to learn Python on the way looking at ML code, but I am probably sticking w/ the compile speed, which is critical for my game engines, and other analysis. I started looking at some lectures of ML and RL, and I think there is a great future for this field that. Thank you for your reply.

I am not *exactly* self-teaching but learning outside of the university model. I rely heavily on books from Amazon, and YouTube.

* It's really hard to objectively know if you've mastered something by just building something for yourself. If you misunderstood a concept, you can just as easily misunderstand it in your code as well. You need external review to help counter that.

* Lots of concepts require a significant application to really use. In my Operating Systems course I built an operating system. That's hard and takes a ton of time even with lots of support / resources. Same with some advanced data structures. Real big data jobs. Etc. If you're self-learning I think its easy to think you've built something that demonstrates the concepts when in reality you didn't even scratch the surface.

* Finally, if all you have is applications, its really hard (aka impossible) to demonstrate to a 3rd party that you actually have mastered a lot of advanced concepts. A 'degree' (even a bootcamp 'degree') is a simple (but very flawed) proxy for what you know.

There's some disconnect here, because I'm planning on being an entrepreneur and being innovative, not prove my worth to an employer.

For peer review, StackExchange and GitHub are options to gain the outside perspective necessary, which I use.

You always have to prove your worth. Literally always. You have to prove it to investors, you have to prove it to the people you want to hire/work with, you have to prove it to your early customers, etc.

Of course, money talks. So you can always prove yourself by making money (but if you're capable of this, then Dave's point stands - stop just 'learning' and start 'doing').

Even given that, my points above still stands. I'd bet something like 95%+ (pulled from ass) of people that have never had external feedback of their learning are vastly over estimating what they think they know. You should seek out actual feedback/review as much as possible.

I'll prove my worth to consumers, and use my shoestrings for investors. I am not going into detail though for obvious reasons.

What's wrong with StackExchange and GitHub?

I mean, sure, but the entire distinction between "self-taught" and ??? seems a bit silly and only matters at the entry level because one is (hopefully, necessarily) learning a lot through the experience of building real software.

To the example of "real big data jobs" -- the cutting edge has always been in practice where the ideas emerged; I would argue that universities don't really scratch the surface of industrial knowledge (or lag it by a few years).

It seems like the wrong dimension to focus on; if one is interested in working in an area which is rapidly changing, one has to learn from the community somehow, but my experience isn't that people who are reading papers, blog posts, source code, etc. to keep up talk about being "self-learners" except to the extent that they want to warn you that there might be some weird holes in their knowledge.

Tons of PhD dissertations are online, and online discussion is everywhere. I'd also consider correspondence after getting farther along.

MIT OpenCourseWare has community for example as do niche forums.

All these challenges can be met w a can-do approach outside of "the box".

Yeah. Make sure you're learning some sales and marketing if you want to make money :-)

I don't imagine that one can get quality code and design feedback on an open source project without some level of adoption so it's a bit chicken and egg, but certainly if one does get that level of adoption/community, that sounds like success.

But that's because something interesting was built.

In machine learning, DeepMind seems quite open about ideas (regularly posts papers on arXiv). Of course it has commercial secrets and the open papers are only the top of the iceberg of research and are lagging, however, there's a decent amount of scientific info about the architecture of learning agents there, which is more important in ML than actual coding, and an entrepreneur can overcome the lag by finding their own niche.

I believe too that sales and marketing, not programming, are the bottleneck of progress on the business path.

The main problem with doing everything on one's own is that it's a high-variance approach. If an interesting and original product is built, it will have no competitors initially because no trade secrets will have been disclosed. If the product turns out not original or functional enough, it will flop completely, with a lot of time and money invested in vain.

By a landslide #1 complaint from folks who freelance that I have talked to is that the selling yourself perpetually part gets really old. Also systems are much more complicated today than in yesteryear and thus a collaborative, highly functional team approach is necessary to build quality systems/products (my opinion). Better be able to convince would be clients that you can add value to the team.