How does one find consulting gigs in machine learning + NLP?

Ask on Quora. Send your resume ;). You're doing fine so far. Another decent place to hang is Hacker News. Meet some of the YCombinator founders and they'll pass along your name if your skills/interests warrant it. Also, submit to http://hirehive.com. Many of the current YC companies are hiring and are using it.

Anon user's "meet people" comment is dead on. Go to where your clients are. Start with familiar territory like user groups and unconferences -- attend, volunteer, present, organize. That's good for some basic networking, but all too often won't provide much in the way of paying work with any frequency or regularity. Write an NLP blog. Write whitepapers. Frequent NLP fora and lists. Become a recognized regional/national/global expert. Advertise. Google text ads do drive traffic when well-written. Sponsoring events can be an effective tactic as well, if they're the right events. Join groups, attend social events, attend business events where potential clients will be. http://Meetup.com events are a great place to meet active savvy professional types. So is the golf course. If you have kids, go to where other kids' parents are. You're probably shooting for upper-middle class/wealthy small business owners, and middle management/hiring managers in larger corps; possibly team leads, but they all too frequently don't want to outsource (for various understandable or irrational reasons) or don't have a lot of influence in bringing in a contractor. Possibly project managers ... There's lots of NLP work to do in the startup world, but rarely a lot of pay, and the smart startups aren't going to outsource their core competencies.

Handle this like any other consultancy. It usually makes the most sense to agree on small concrete tasks with measurable performance and accuracy. Usually the best places to go are Big Data meetups. Many of the people that go to those events have just started to collect large amounts of data but don't know what to do. Listen to their needs and see if you can offer them something. Andrew's advice is the most spot on. In addition to his, make a demo website that highlights what your software can do.

Many companies do not want machine learning consultants because you are working on the crown jewels--and they want to own this technology. In particular,  avoid startups--startups want employees.  Go to really, really large organizations.  Frequently they hire consultants because it is much less trouble to get rid of a consultant to fire an employee. A startup does not care.  Indeed, they would rather fire you early, learn all you know, and keep your options for themselves.    At a large company, the employee can take credit for your work, and blame you if something goes wrong.  Also, there is so much going on, your coming or going is not going to break the company. The best place to get start is what I call a "sunshine lab".  Big companies set R&D labs under the premise of innovation--of course, many just can not innovate and really these operations exists solely for political reasons.  This is an ideal place to start because you can just hang out , get some real world experience.  You then have time to do your own stuff, learn lots of coding tricks, think about a startup, meet lots of people, etc.  Maybe you will have a minor impact on something; you may be surprised incredibly how hard it is to get even 10 lines of code into production, even when everyone is trying really hard, at a really big firm. Timing is also important.  Frequently it is easier to get consulting contracts mid year (note I mean fiscal year).  If a manager does not spend their budget for this quarter or year, then they won't get budget for next year. Expect a delay.  Frequently companies will search for an employee first.  They might contact you and inquire.  Let them know politely you are available for consulting, not FTE.  They will then look around, maybe for 3 months or longer.  If they don't find anyone, then they will come back to you.  It is better if you can align yourself with an existing firm that provides consultant staffing.  Frequently they might provide 100 IT contractors, and therefore have the channel to get into the meatier work of machine learning.  Many large companies use tier firms that hold the contact (and the insurance).  Also, many large firms may not advertise the machine learning positions--you have to inquire and see if the staffing firms know of a project that could use an expert.  I actually got my first consulting gig by coming in as a computational chemist / java GUI programmer, and met a VP in another division wanting to try some machine learning.     I had worked at a startup years prior doing machine learning, and with the PhD and the pedigree, I had the chops to get it rolling. Once you have some experience consulting, this is what you are selling--knowing how to get things done in a big place.  The machine learning is almost secondary.