My skillset when I started consulting in 2019 was primarily deep learning for vision and NLP.

In 2020, it expanded: neural search; recommendations; predictive analytics beyond deep learning; and ML for satellite data.

But these skills were mostly reactive to client needs.

I've always been a fan of @woj_zaremba's 'Learning Day' at @OpenAI. Everyone spends a day a week learning skills beyond their day job.

But I struggled to adopt it myself since my income was directly tied to time spent on client projects (hourly billing). https://openai.com/blog/learning-day/

This quarter I started transitioning from hourly billing to biweekly retainers. I can finally afford some schedule flexibility for something like the Learning Day.

@patio11 + @tqbf's advice on HN and elsewhere was extremely helpful. One of my favorites:

https://news.ycombinator.com/item?id=4247615 

In 2021, I'm looking forward to focusing my initial Learning Days on: low/no-code; MLSys and MLOps; and ML for pricing and demand optimization.

I'm confident these skills will help make my data science and ML practice substantially more impactful.

Exciting year ahead!

Interesting thread corroborating my decision to focus on machine learning for pricing: https://twitter.com/trengriffin/status/1344694148931018756