Every year I set myself a theme and 2023 was no different: The Year of Focus.

But secretly it turned out to be the year of AI.

ChatGPT taking off like it did and with it an entire boom around large language models.

Overwhelming to say the least. But beyond LLMs, I talked to many scientists and they have seen machine learning-driven disruptive innovation in their respective fields as well.

It was a year of not just dreaming big but also achieving big, a time when my passion for Earth sciences and artificial intelligence converged to redefine the boundaries of what's possible.

With GraphCast neatly bundled under our collective ArXiv-Christmas trees, 2023 shaped up to be a frenzy of learning and building.

Innovating Weather Forecasting with Machine Learning

In 2023, my primary focus was partaking in the revolution of weather forecasting at ECMWF.

I was lucky enough to take a part in integrating advanced machine learning algorithms with traditional meteorological data.

Many beautiful developments in weather prediction. This innovation will not only provide faster and more reliable forecasts but also played a crucial role in enabling rapid experiments and innovation in these models for non-machine learning experts.

The rise of machine learning in weather forecasting marked the beginning of our public development of a ECMWF-owned fully data-driven numerical weather prediction system.

But we also open-sourced a solution to run these novel models, like FourCastNet and GraphCast.

We on-boarded new colleagues, built out robustness, and operationalised these models for meteorologists to evaluate data-driven NWP on their own terms.

What an exciting year it was!

Diversifying Machine Learning Applications

Beyond weather forecasting, I explored the reproducibility of machine learning in science and beyond.

In my role as a fellow of the Software Sustainability Institute, I championed the cause of sustainable coding practices in machine learning.

Technically, my SSI fellowship was for 2022, but I still had some life in me and published an invaluable resource at the beginning of the year.


This niche resource attracts 100s of scientists every month who can improve their own research with these "quick wins" to improve their ML-based science innovation.

By focusing on reproducibility and efficient coding techniques, I aimed to set new standards in software development, particularly in the context of AI research.

I carried this into my work with the ITU Focus Group on AI 4 Natural Disaster Management and by publishing Data-Science-Gui.de.

Educating and Inspiring Through Digital Platforms

As a Youtube Partner and Skillshare teacher, I dedicated substantial effort to demystifying machine learning concepts for a broader audience.

My tutorials and classes, ranging from data science fundamentals to the intricacies of stable diffusion and chatGPT, garnered significant attention, helping to nurture a community of aspiring ML practitioners.

I published a course about chatGPT for creatives, which ended up being my fastest-growing course ever published.

But I didn't end there...

I also published the book ChatGPT for Creative and Content Creators. A 36-page explanation of what chatGPT and AI is and how to properly use prompt engineering for content creation.

Very proud of these achievements!

While I didn't want to give any presentations this year, in the futile attempt to "focus", I was invited to give a guest lecture at Brown University and traveled to the Netherlands to speak about operationalizing ML in weather forecasting.

And all the while, I still wrote my weekly newsletter that takes AI and machine learning beyond the hype to over 1111 subscribers.


As I reflect on the whirlwind that was 2023, I'm filled with a sense of accomplishment and anticipation for what the future holds.

I was lucky to be a part of some amazing teams and interacted with lovely individuals across organisations to build at the very leading edge of machine learning and AI in multiple disciplines.

As we stand at the cusp of another year brimming with potential, I'm left pondering – what new frontiers of machine learning will we explore next?

PS: 2023 was also the year I dove with sharks, who such fascinating creatures!

Bonus picture for those that are still reading:

Jesper Dramsch diving with sharks