I'm not sure why OP thinks this is going to be a good idea. The kind of projects he'll get to work on will almost always be the usual Odesk "clone XYZ in $X" tripe. Also, the sort of company he's talking about is unlikely to have senior developers experienced enough to mentor him (from what I know about these companies.) Working on an open source project instead would be much more constructive and would get him engaged with experienced developers who could teach him a thing or two. If he's still a student (and it looks like he is) GSoC might be a great way to get started with this.
Also, I would avoid Odesk like the plague. Build up a portfolio (open source works as well) and connect with people on HN.
> The kind of projects he'll get to work on will almost always be the usual Odesk "clone XYZ in $X" tripe.
He knows that much and noted it.
What he wants to learn is discipline and getting shit done no matter how much it smells. Having a boss breathing down his neck might work.
> Working on an open source project instead would be much more constructive and would get him engaged with experienced developers who could teach him a thing or two. If he's still a student (and it looks like he is) GSoC might be a great way to get started with this.
No. In an OSS contribution (or most GSoC) if you lose yourself nobody will care much, it's your loss. You're completely missing the point.
I did some projects from both oDesk and HN. They were quite similar. oDesk projects I've chosen were more interesting but quite short and HN jobs were longer but not so great as you would expect from HN.
How do you go about building a portfolio? Do you do free/cheap work for people, or do you just make things you think people might like or be impressed by?
Also, I would avoid Odesk like the plague. Build up a portfolio (open source works as well) and connect with people on HN.