First and foremost, credentials for CS related AI and machine learning are largely meaningless right now.
So my first piece of advice is to quit 'seeking' instruction like a computer waiting for further input and get into motivating yourself through self study.
Secondly, keep in mind that machine learning is something that will take an enormous amount of time out of your schedule after you've gotten the basics of the learning engine completed. You HAVE to interact with it and allow others to in order for it to truly learn.
With that said, here's some wonderful ways to help others with their AI projects - and also a way for you to to get started researching and studying, from the outside, the dynamics of dialog and interaction that you'll be working on as a CS programmer.
Elbot: http://elbot_e.csoica.artifici...
Cleverbot: http://www.cleverbot.com/
Existor (her name's Evie) is based on the cleverbot script: https://www.existor.com/en/
and Skynet: http://www.skynet-ai.com/
Third. You're an engineer by trade. If you truly want to understand how to make a machine think. Then take psychology courses, marketing courses, education courses, economics beyond macro and micro are all helpful to understand psychological motivation of populations, and more. Why do all this? A machine can 'wait' and consume information, but that doesn't make it intelligent. What makes it intelligent is it's desire to participate in the community it belongs to and that belongs to it. Psychology - whether it's through market forces or internalized - is what we now know as a population motivates. Integrating these into an AI is critical.
Fourth. Take a look information storage and retrieval systems and become an expert in databases, weighted algorithms, and different levels of normalization. The book 'Data Insights' By Hunter Whitney is a wonderful book on information systems and the different potential ways to perceive data. If you're poor like I am, Hunter has distributed a full copy of Data Insights through torrent web sites, with his only request being: If you can afford it, and the book has provided benefit to you, then please pay for the real copy. you can find at any Barnes and Noble in the country
This leads directly to neural networking. My advice from there is to dig into peer to peer networking and to understand how these systems function. Bitcoin's open source, and provides a wonderful example of what not to do with a peer to peer network and information storage, which you can see by the massive gigabit chain you have to download.
Why this is all necessary:
With a MS in CS and 15 years experience, you should by now be able to create at least a mid sized client server or n-tier application, end to end.
Now you gotta figure out your input stimulus for your AI. Are you acquiring information from text input alone? Are you acquiring it through a Kinect device connected via a USB and pulling out 3d data and sound? Are you placing your AI on the internet as a chatbot? Will the thing be mobile? If so, how?
Knowing your stimulus and nailing it down to a few input devices is crucial to developing a learning system.
From there, your next goal is to develop the support systems which 'go' with the AI.
And this can WILDLY vary depending on your methods of stimulation.
For the most part though, if you don't have proficiency with databases and data stores, Then you're not going to understand memory retention schemes for AI properly and how and when to optimize your database and the differences in normalization schemes.
So go get a job in databases for a few years then come back. These are a dime a dozen and easy to find anywhere. Pick your database wisely, you'll probably stick with it for your career - and it's hard not to be a database bigot afterwards.
If you don't have proficiency in middle tier type work or embedded systems which might leverage various forms of publisher/subscribe methods under load and real time condition, then get a job with a real time mission critical information group. Flight Systems such as Boeing, Rocket Systems Orbital (my former employer in Chandler, Arizona) are excellent ways to get experience - albeit very stressful work - for highly responsive AIs. But if your AI doesn't need to be absolutely real time, then you can always get a job which specializes in batch processing. Financial companies such as Wells Fargo (another former employer of mine), payroll systems such as ADP, or any billing center / general ledger operation will show you the benefits and necessities of synchronizing distributed transactions.
Finally. Like a corporate application or web presentation, there's a UI and presentation layer for machine learning. It's different in that fact that you're trying to make the interface proactive rather than reactive. We as humans need that stimulus and machine learning is predicated on receiving feedback as well. So if you can't provide a sufficient enough user experience for the machine, then the user interest's level wanes.
This requires design experience. So if you have never done User Interface design work and worked directly with customers and QA staff, then my advice is take a job programming in one of these areas for a bit - preferably with a great deal of customer interaction where YOU ARE the programmer and analyst.
Now keep in mind if you pursue education with machine learning in a structured form, you're limiting yourself to the concepts and ideas and imagination of the instructor and material of the presenter.
Machine learning is something you teach yourself.
Until you do. Then you're the machine that's learning. So quit projecting.