I’ve had a programming blog tumblr that I started way back in 2015 before attending App Academy – it was a requirement at a/A to publish a post every day – this would not only show what we’d learned that day but also help to solidify the knowledge we were imbibing from the proverbial firehose. I continued to sporadically post to this blog even after getting my first (and so far only) programming job.

Now I no longer have a job. I quit a few months ago to travel in Nepal. I also ended up traveling to India. I spent half of my time in India studying machine learning at Hoztel Jaipur. The owner of that hostel had quit his job in tech to build a data-driven chain of hostels in India – we enjoyed several conversations radically different from the standard “where have you been and where are you going” traveling tropes – he suggested I read about Bayesian Learning and Markov Chains. I mostly read about more basic topics from Machine Learning: An Algorithmic Perspective and watched Khan Academy videos on gradients and partial derivatives. At one point my mom asked me “but why do you have to do this in India?” I had just gotten mild food poisoning and at that moment didn’t have a legitimate answer and so that very same day booked a ticket back to the States. I spent my last few days in India finally touring the magnificient forts of Jaipur and the profound and breathtaking and tear-inducing Taj Mahal in Agra.

At a/A they suggested we use tumblr – so as to focus on writing the blog, rather than on maintaining one. I’m fairly prone to getting lost in the weeds of side-tasks, so I tend to have to remind myself that the point is to write about my progress and not to learn how to build a super-awesome blog. Or that the point is to learn about data science and machine learning and not to learn Python. Or that the point is to become a more well-rounded engineer and not just to learn about ML. I do see the constellation of data science skills as a complementary addition to my existing software engineering skills. I imagine that I will not so much “become a machine learning engineer” as “learn machine learning while continuing to be an engineer.” Or that the point is to work on interesting shit and to grow and to not stagnate and not so much to “learn machine learning.” That said, ML is a field I’ve been interested in and dancing around for quite some time. Abstractly it feels like the underpinning of much of the current progress in our civilization; concretely it is a challenging field to break into due to its interdisciplinary nature. It’s as good a reason as any to study mathematics – and the more I study mathematics the more I realize how fundamental it is to progress in our civilization. In the past I’d thought about feeding all of Mahler through some sort of neural network so that it would create more music like Mahler. An exciting proposition in theory. Being jobless, I have the opportunity to give it my all (not to the Mahler thing, but to the whole learning project).

The Python ecosystem provides many tools and frameworks for data science and machine learning (and deep learning) and so I will have to learn to better wield Python (I had worked a bit with it in my last job, as a result, in fact, of working on projects with a data science team) as well as NumPy, pandas, and associated graphing toolchains (matplotlib/Plotly/Seaborn). But of course the point is not to just learn these tools – I’ll also be learning machine learning theory – like how linear and logistic regression work, what exactly are perceptrons, backpropagation, neural networks, supervised and unsupervised learning and other modeling techniques. And then to learn ML theory it is necessary to ultimately reach down into the mathematics which is its foundation – in fact machine learning seems to me now to be nothing more than a creative composition of disparate topics in mathematics: statistics, probability, linear algebra, and of course calculus (which, by the way, research shows had been discovered in India well before the time of Leibnitz and Newton). And finally there ought really to be some practical upside to all this – some sort of projects. Ultimately there will need to be because I can’t just study forever. For now I can, and I will.

So as a first step, I found this 1,2,3 guide to publishing via Github pages. I really want to have a simple way to post blogs and to own the content. I don’t care that the content is completely public – that’s an additional point of the blog – it will ultimately contain writeups of little data science projects I’ll be working on that will prove to future employers that these are skills I’ll be able to bring to the table.

Onwards, to engineering, mathematics, and predictions. It’s an exciting time.