Getting on Track: WindyCity Rails Presentations and Presenters that Reaffirmed My Choice to Pursue the Craft
Last monday was my first day as a resident apprentice at 8th Light. I was fortunate to have started the same week as WindyCity Rails, a conference produced by 8th Light and ChicagoRuby. I was able to attend and help with the set-up and day-to-day operation during the conference, which was exciting in its own right because it was my first software ‘developer’ conference, and Rails is the framework I’ve spent the most time in; having graduated this past May from Code Platoon, a non-profit coding bootcamp for veterans here in Chicago. Rails is a core component of the curriculuum.
For me, it was a happy surprise to find that the speakers were not limited to discussing the Rails framework; in fact the presentations I connected most with, had scope, relevancy, and application at least industry-wide. In this series I’ll highlight 3 presentations that consumed me as I consumed them, and explain why they each helped re-affirm my choice and strengthen my resolve to become a hard-nosed crafter of software in a community of professionals.
-
The Hitchhiker’s Guide to Machine Learning Randall Thomas, Thunderbolt Labs
I’ve recently developed a small obsession with Machine Learning. During the lightning talk portion of my 8th Light interview I chose to show off a small application I hacked together that when run would cause my MacBook to ‘magically’ begin describing aloud labels it assigned to real-world objects placed in front of the computer. This magic was smoke and mirrors: A creative but simple Ruby program, using backticks to send system commands that made my Mac snap a secret photo, and speak aloud what was in the photo. It labeled the objects by parsing a JSON response it received back after sending the photo to the Google Cloud Vision API. Simple, right? Perhaps, if you don’t count the massive amount of hidden complexity where enigmatic Machine Learning algorithms that were trained on unfahtomable swaths of image data from which a highly complex model formed enabling an animal like ability to ‘see’ and identify objects.
In Randall’s presentation he illuminated and reinforced this complexity, cartographing his recommendation for a rigorous path by which one may begin to actually understand how we teach computers to learn. What I liked most about this presentation was that he didn’t speak of how challenging the ML path to understanding was in an attempt to scare-off interested but naive application builders like myself, quite the opposite he urged us toward the challenge. Randall poignantly described through historical example, the dangers of runaway technology in the hands of malicious or uninformed actors, and why decent people ought not be timid, and instead be vigilant. Somehow he did all of that and still managed to throw in some solid zingers… I recall at least one tasteful F-bomb.
An hour or so after he left the stage I found him getting mic’d up for an interview and quickly introduced myself and thanked him for his talk. “Nice meatball!” He said pointing at my NASA hat. I feel like a peon in this industry so when brilliant well known people are approachable and genuinely nice to me it really makes me believe in the future of empathy in the software community. I asked him, “So, I’m curious, why didn’t you mention Tensorflow or the Google Cloud Machine Learning APIs?” I’m paraphrasing but he reconfirmed that even though one could devise practical applications and make an attempt at using tools like I had during my aforementioned lightning talk, they wouldn’t be empowered to effectively weild, manipulate, or refine the math that is the basis for any machine learning system. “Plus,” he said, “…you might spend 9 months learning Tensorflow versus actually learning the foundational components of machine learning which would get you up and running with Tensorflow quickly anyway.” I thanked him for the advice and he explained how to reach out if I had any other questions.
He probably saved me months on what would’ve been a less than fruitful pursuit of a single ML framework, additionally having a non-judgmental interaction like that with a veteran engineer makes me want to in turn pay kindness forward and indeed stay grounded as I lurch toward a familiariaty with convulluted topics and comfortability with complex systems. This thoughtful approach to colleagues, regardless of which rung they sit on the ladder to mastery, is in part how we might rise the tide and lift all the boats, creating a community of professionals.
Check out Randall’s ML Recommendations & Resources: https://gist.github.com/daksis/a00816eb5149920b266b3758e3823542