This year, I hoped I could write my first-year review before it ends, but I delayed(, life!). I want to take a step back and see what went well, what I could have done better, and what I look forward to in 2022!1.
This review is a reflection of my own journey. I am writing it to serve as a reference of where I came from and what I aspire to achieve going forward. This review is centered around my important values which are being useful to others in form of what I create and share, community, and learning actively. I am going to reflect on what went well across those values, what I did not do well, and what I look forward to in 2022.
What went well in 2021?
1. Creating content
At the beginning of 2021, I started to create content on Twitter and LinkedIn. Later, I started writing weekly articles on Hashnode and Medium. I grew my Twitter audience from a few hundred followers to 7500+ followers(as of writing this article).
One of the things that I am very happy about is the complete machine learning package that I published in September. The package contains over 32 end-to-end and interactive notebooks on different machine learning topics and tools. It has been starred over 2000 times. But the most important thing for me is not the number of stars or folks, it’s the kind messages or comments that I receive from people that used it to practice or refresh their machine learning skills.
On average, for the past three months, my tweets and threads got 3.5M impressions and around 35K impressions per day. Also, some of my tweets such as this and this were shared by remarkable people in the machine learning community such as Yann LeCun, Francois Chollet, Nando de Freitas, my friend Sebastian and many more others, to name a few. Andrej Karpathy also reacted to one of my computer vision tweets that were about the fact that vision is a hard problem and shared that he is going to write about the state of computer vision following up on his 2012 blog post. I look forward to reading his views on the state of computer vision in October 2022. My content doesn’t deserve to be shared by such amazing people, so, I am very grateful and humbled by their supports!
I also wrote 27 articles on Medium and Hashnode about different machine learning ideas and techniques.
These are what went well about my writing and content creation in general. Overall, I achieved a lot, but there are also lots of things to learn. For example, I had not been writing weekly articles but I am planning to resume it, and hopefully, this review will give me a head start.
Earlier this year, I was eager to learn about machine learning operations(MLOps). It is reasonably easy to build machine learning models, but it’s very difficult to put those models in production. There are lots of things to take care of, and also, models are naturally prone to changes. Data changes too. The fact that data and models drift over time makes it extremely hard to build machine learning systems that can work reliably in a production setting for a prolonged period of time.
My goal was to get a reasonable understanding of it so that I can later dive deep into it. To learn it, I read the book Introducing MLOps and I took the first 3 courses of Machine Learning Engineering for Production (MLOps) Specialization by DeepLearning.AI.
Most of my learning in 2021 happened during the time I was designing the complete machine learning package. I can’t say how much it helped me to understand things that I thought I knew, but I actually didn’t. Of course, I learned other things from events and my favorite blog posts and other creators.
Being in a fast-evolving field like machine learning means learning consistently. I still struggle to keep a good learning habit, but my goal this year is to figure out how I can sustain my learning. There are lots of cool things hanging around that I would like to learn, and I will always share my progress in articles or on Twitter.
I got in contact with machine learning from attending community meetups. I thus make it my goal to participate in communities that I care about the most as much as I can.
Also, as Event Ambassador at DeepLearning.AI, I managed to host 2 events about learning machine learning with Santiago Valdarrama and A vision case study in self-driving cars with Vladimir Haltakov. In 2022, I am planning to make more video content, and you can now subscribe to my Youtube channel in the meantime.
Occasionally, I like to hang out on tool forums to know what’s new with my primary tools.
### What did not go well
Overall, I had a great year. But also, as I said before, not everything went well and there are definitely lots of things to learn.
1. Being consistent
For about 2-3 months, I wrote weekly articles consistently but suddenly stopped writing regularly. Writing articles was becoming one of my favorite activities, and I am certainly going to resume again.
2. Unfulfilled promises
I said publicly that the first issue of my Deep Learning Revision newsletter would be out in December, but I didn’t accomplish it. I am definitely going to act on it. Different to what I said, the newsletter will be weekly rather than monthly and it will be all about the ideas and the latest advances in deep learning for computer vision and where vision intersects with language. You can subscribe to the newsletter here!
What I look forward to in 2022
My central goal this year is to learn engineering practices related to machine learning. I am also going to challenge myself to make videos, write regularly, and keep showing up on Twitter with useful content.
Thanks for reading!
Each week, with a probability of 80%, I write one article about machine learning techniques, ideas, or best practices.
P.S: I just finished college 🕺🏻
Thanks to James Clear for sharing his annual review format. It helped me to organize my own review! ↩