2017 Review
It’s time to review what happened in 2017.
I completed the new version of fast.ai Deep Learning course. I attended it in person at the University of San Francisco; it was a great learning experience. This course is very helpful for the understanding of how to use deep learning in practice. There are tons of little tricks, so-called “black magic of deep learning,” that is not covered by textbooks or typical ML courses in universities or on MOOC platforms. This course makes emphasis on uncovering the magic of deep learning and “making neural networks uncool again.”
I went to a few ML-related conferences. The largest was NVIDIA GPU Technology Conferenc (Silicon Valley). AI By The Bay was interesting. Smaller ones were good too: TensorBeat, AI Tech Forum. And one Python dev conference: PyBay.
Also, went to a countless number of meetups. The best meetups were “Engineering Leadership talks by Facebook Executives” and a series of ML engineering meetups at Twitter HQ.
Continued to lead a team that works on chatbots at JustAnswer but shifted focus from product development to chatbot development tools and underlying technology.
ML projects were mostly focused on text classification and natural language understanding for chatbots: intent classification, named entity recognition.
Got more experience with communicating different aspects of machine learning to various small and large (up to 100 persons) groups: engineers, product managers, engineering managers, line-of-business managers, executives.
Started working with contextual bandit algorithms. Published an article about relationship between contextual bandits and reinforcement learning. In practice, contextual bandits are tricky to train; there are quite a few subtle issues that can arise if you are not prepared.
I wrote 17 blog posts (almost double from 9 posts in 2016). It was a mix of short posts that didn’t get much attention and a few long posts that got thousands of views, probably because I was not embarrassed to advertise them. These 17 posts got about 26k views on Medium and a few thousand on pavel.surmenok.com.
At the end of the year I got lazy and stopped cross-posting to pavel.surmenok.com, so my latest articles are only on Medium. Overall, I get about 13k unique monthly users on two platforms combined.
I planned to grow the number of followers on Medium to 1000 but overshot the goal and got 1338 followers by the end of 2017.
Medium called me a “top writer in Artificial Intelligence” and (surprisingly for me) a “top writer in Self Driving Cars.” I lost both badges by now, need to resume posting more related stuff to get to back to the top :)
The most popular articles in 2017:
I changed home location a few months ago: moved from San Rafael (North Bay) to Redwood City (Peninsula). So, I’m still in the San Francisco Bay Area, just more to the South now.
Continued listening a few ML related podcasts and, occasionally, some podcasts about startups and technology. Started listening “Masters of Scale” by Reid Hoffman: many thought-provoking topics and great guests (Reed Hastings, Peter Thiel, Mark Zuckerberg, Eric Schmidt, Sheryl Sandberg, to name a few).
Some of the books I read (or started reading):
This is the best software engineering book I’ve read in the last few years. It is a dense and deep guide to building reliable and scalable distributed systems. Authors describe different aspects of the software systems, what can possibly go wrong, how to make sure the system can be available and function correctly in case of faults in parts of the system. Everybody who designs and develops distributed systems will benefit from this book.
- “The Start-up of You: Adapt to the Future, Invest in Yourself, and Transform Your Career” by Reid Hoffman
This book has an interesting perspective on managing a career using the same principles as entrepreneurs manage startups. In addition to theoretical constructs and real-life examples, the authors provide actionable advice.
- “Deep Learning” by Ian Goodfellow.
It’s a dense introduction to a broad number of topics in deep learning, mathematical background, deep learning techniques used in industry, and research perspectives.
- “WTF?: What’s the Future and Why It’s Up to Us” by Tim O’Reilly.
Tim O’Reilly has a long history of accurate future predictions in tech. In this book, he shares his approach to predicting the future.
How was your 2017?
