2018 is looking busy, but there’s still a lot left from 2017 that I didn’t write about or wrap up. I’ve been a bad blogger.
I finished the Udacity Nanodegree towards the end of the year, with a bit of work-induced delay.
There isn’t a lot to write about after my last update that isn’t pretty much covered by the syllabus. As noted, I had to take a bit of a break in August and September so I finished a bit late. In the end, I was happy with the level of instruction and the knowledge I gained.
To sum up, the class was a lot more “how” than “why?” or “when?” So while I feel fairly proficient at using TensorFlow, I’m somewhat less conversant in when it would be best to use the various techniques to solve real-world problems. An effort to sharpen my skills and become more comfortable with when to use (and not to use!) the various techniques has consumed me since then. I continue to think this is important to my current focus on geospatial and mapping applications, but it’s also become clear that there are a lot of other simpler tools that could be used for a lot of the things I’m trying to do. As a person I follow on twitter noted recently, the hardest part of any of this is collecting and figuring out the data. After that, often the tools needed are far simpler than TensorFlow and neural nets. The key is knowing the difference. That takes a lot longer than learning the tool.
Transportation Data Challenge
The National Transportation Data Challenge wrapped up in November. I was not able to make it to the summit event that concluded it, as it was in Washington and I had other things going on here. A lot of good stuff was accomplished and presented, about which I’ll be speaking in a couple of weeks (see below). However, the end sort of left a bitter taste to me, in that it seemed to just stop. There was not even an update to the website when it was all done, let alone a link to any videos or accomplishment. As a project manager who’s been taught to always wrap things up neatly, it seems somehow unfinished.
I had a fun time at PyCascades in Vancouver a few weeks ago. Unfortunately I hit a week that was unusually rainy even by Vancouver-in-January standards. Fortunately, I was there for the Vancouver Hot Chocolate Festival. That worked out pretty well because as is the case with many single-track conferences, I found myself with some free time on my hands during sessions that I wasn’t particularly interested in. The reality is that while I’m interested in using Python for a lot of things, I always need to remember that with a general-purpose and extendable language like Python, there will always be people doing things I’m not. For example, I’d rather shoot myself than build any front-end UI type stuff. But there are lots of people using Django and other frameworks for websites and I’m pretty much clueless sitting in those sessions. I probably should at some point familiarize myself.
I’ll be speaking at SCaLE 16x in Pasadena in two weeks. The discount code is “SPEAK” and it gets you half off the ridiculously low regular price of $85. It’s a four day conference, that’s got to be worth it! My talk is titled Open Data Initiatives: what works, what doesn’t, and what you can expect. It somewhat leverages my experience with the National Transportation Data Challenge last year. I posted my promo video for this one yesterday. It includes cameos by Mr. Spock and Flash the Linux Cat.
I also submitted a proposed talk for the education talk at ITX-NZ in Wellington in July. Sigh, another excuse for a winter trip to New Zealand. Maybe I’ll get in shape and ski this time, though that’s actually a bit early in the season. It addresses what I see as challenges in how we teach computer science and engineering skills in the current era. Again, somewhat inspired by a twitter conversation in which several people were lamenting basic skills in people who had superb knowledge of advanced topics. It seems to me that if we want to force people to master basic skills like memory management and efficient programming, we need to force them to spend a lot more of their education on simpler infrastructure. The talk is titled “Can you do that on a Raspberry Pi?” and if not accepted at ITX, perhaps I’ll present it elsewhere.
If ITX doesn’t happen, I will likely go to either PyCon US in May or SciPy in July (same time as ITX). I’m going to try to avoid spending too much time and money traveling to conferences out of the area this year, and more time looking for work doing this stuff.
Work, work, work, work…
Had a couple of major project come up in January, one of which has dragged on a bit. Nothing terribly interesting, but it’s money and the clients are people I’ve worked for before. It is a reminder that I need to spend more time looking for data-related work rather than ongoing IT work with existing clients.
I’m mostly keeping my personal stuff over at adv.michaelgat.com, my photo/adventure blog. Some photos may migrate over here, but that’s where my 365 day photo project and my occasional stargazing and other adventures reside. Also, microadventures, Griffith the Traveling Mountain Lion and an eclipse!