2022 was a peaceful year for me. While it’s been an anxious 4th quarter for tech, especially if you were affected by layoffs, the overall year has been kind to my family and me.
Now that Latte’s grown up (18 months old), my wife and I got to travel more. First, back to Singapore, then Vancouver, Hawaii, and New York. We also explored Washington (e.g., Mount Vernon, San Juan Islands) and hiked a lot over summer. And despite some of us catching Covid, my immediate and extended family continues to stay healthy.
Write 26 posts: Partially completed (18/26). While it wasn’t a deliberate goal, I found myself focusing on topics that fit the Start Here page, especially technical subjects such as bandits, counterfactual evaluation, design patterns, Python, pipeline testing, text-to-image, etc. I learned a lot while writing them and got feedback that they were helpful.
Work with an editor: Not started. I procrastinated as I wasn’t sure how to work with an editor on my writing, especially the more technical pieces.
Learn something new: Completed. I had the opportunity to learn about and apply bandits, counterfactual evaluation, text-to-image, pipeline testing, etc. at work. Also, while in Hawaii, my wife and I learned to surf!
3 mentees reaching the next level: Partially completed (2/3). One mentee got promoted to a lead role with direct reports while another switched orgs to lead a new team.
Learn to snowboard: Not started. Couldn’t get away at the start of the year.
Learn Julia: Deprioritized to focus on machine learning techniques, engineering, and PyTorch. Will probably not pursue unless there’s increased industry adoption.
Continue angel investing: Completed. I’ve been getting feedback on how my previous investments did and refined my investment thesis on founding team, TAM, market pull, etc. While I continued investing in 2022, it’s reduced relative to previous years.
Promoted from L5 to L6 in Q1. I felt that I was performing at the L6 level since early 2021 and it was reassuring to be recognized for it. This comes with more responsibility and work and I welcome the opportunity to stretch. One of my goals in 2021 was to focus more on ML strategy and product; I think working on this helped with the promo, and becoming an L6 opens doors to do more of it. Excited to focus more on it.
Through this process, I learned that excessive desire can be unhealthy. I had been feeling under-leveled since early 2021 and needed the increase in title (and compensation) to match my performance. This wastefully consumed emotional energy and caused a few sleepless nights. Looking back, the desire and emotion wasn’t worth it: I don’t think it accelerated the process and just caused unnecessary strain on myself.
I also learned to focus on internal rather than external scorecards. To paraphrase Warren Buffet: Would you rather have L7 performance but an L6 title? Or would you rather have L6 performance but an (undeserved) L7 title? Ideally, they should align but if I had to pick, I rather have the former and then work towards the external scorecard catching up; I’m not a “fake it till you make it” person (though I’ll speak up or walk away if it’s unreasonable).
Gave a keynote at RecSys. After speaking at 10 events in 2021, I wanted a break in 2022. Nonetheless, I accepted the invitation to give a keynote at the RecSys workshop for online recommender systems because it was held in-person in Seattle. Also, being an invited speaker, I could talk about anything(!) without needing to submit a proposal.
For the talk, I decided to persuade the audience that they don’t need an online (aka real-time) recommender system for most use cases; batch is good enough. (Blasphemy, given that this was the keynote for the online recommender systems workshop. But I’ve preached this enough times to make it a talk and not repeat myself.) Then, I focused on scenarios where online recommender systems are more effective than batch–where the juice is worth the squeeze–and shared three examples from my team at Amazon.
Based on the positive messages received, people seemed to like the keynote. I also enjoyed the process of preparing for it and sharing my team’s work. Might do another in 2023.
Leaving this here for when imposter syndrome inevitably strikes again. pic.twitter.com/Oe9VAdsG1W— Eugene Yan (@eugeneyan) September 25, 2022
Started an online machine learning meetup. Some statistics on ML Meetups Virtual:
We now have a small team in place to sustainably host meetups once a month in 2023. Excited to attend more of these!
With the goals and misses of 2022 in mind, here are my goals for 2023.
Write 26 posts: No change from 2022. I’ll continue focusing on writing that fits on the Start Here page, such as machine learning, engineering, mechanisms, and learning. As a sub-goal, I’ll aim to work with an editor for ≥ 4 pieces.
I also want to give myself space to write about other topics such as people, productivity, and product, even if no one’s interested in my take. There’s plenty of wisdom to be distilled from my conversations with mentors and mentees. Thus, as a sub-goal, I’ll aim to write four of these smaller (~1,000 words), column-like pieces.
Learn something new: In 2023, I’m keen to dive deeper into the fundamentals of search, and work on my skills in Python, engineering, and MLOps. (The latter shows up in my 2022 writing on design patterns, Python, pipeline testing, etc. I enjoyed them and will likely write more on these subjects in 2023.) I also want to learn more about business and product so I can better support the overall org. This adds two more writing sub-goals: Four posts on engineering and four on business/product.
I’ll also work on getting into circles where I’m the dumbest person in the room. (Currently, I’m invited into rooms as a subject matter expert and we mostly discuss topics I’m familiar with.) There are some good suggestions here and would love your advice if you have any.
Work with a career coach: I learned and got feedback that I push too hard and take failure harder than most. I need to learn how to better manage this for a forty-year career.
Advance the industry: This can take several forms such as writing useful posts, giving good talks, mentoring and helping others grow, organizing meetups to share knowledge, etc.
Learn to snowboard: With Latte grown up, it might be easier to leave her with a friend while we take weekend snowboarding lessons.
Read more fiction: The bulk of my reading is non-fiction. Of the fiction I’ve read, I enjoy sci-fi, alternate history, and short stories (e.g., Exhalation). Recommendations welcome! Here are my favorite reads this year:
Meditate 60 minutes every morning: I neglected this while taking care of Latte in the morning, or while trying to get a head start on work/writing.
Total pageviews on this site grew from 337k to 436k (29%) driven by 211k users.
The top pages in 2022 were a mix of technical (e.g., python, design patterns, recsys design) and non-technical writing (e.g., why-what-how, onboarding, simplicity).
Clicks via Google Search increased from 50k to 77k (+54%) while impressions increased from 1.52M to 2.14M (+40%). CTR increased from 3.3% to 3.6% despite average position dropping from 27.2 to 29.6.
Email subscriber count grew from 2.5k to 4.2k. Open rate continues to be decent at 65% though I expect it to decrease as earlier subscribers start to drop out.
Twitter followers grew from 6.5k to 9.5k while LinkedIn followers grew from 8.8k to 27.1k.
That’s it for my 2022 year in review. I hope you had a great year too. If you’re writing a review or reflection, please send it my way! Love to hear what y’all are working on and have been over the years.
Thanks to Yang Xinyi for reading drafts of this.
Join 4,300+ readers getting updates on data science, data/ML systems, and career.
Welcome gift: 5-day email course on How to be an Effective Data Scientist 🚀