Dasha Herrmannova

💡Final thought: Improving reproducibility isn’t just about following best practices—it’s about building trust in the results we share and ensuring that our work stands the test of time.

There are a lot more insights in the paper that won't fit in this post. If you care about reproducibility, I can highly recommend giving it a read.

Improving Reproducibility of Gen AI Evaluations Would you trust results that you can’t reproduce? Reproducibility is the backbone of trust in AI research. Without it, we risk misleading conclusions, wasted effort, and barriers to meaningful innovation.

I think this is one of the most important skills to master as a data scientist. Do you agree?

In my last post, I shared why I believe learning NLP is a smart investment—even as generative AI takes center stage. Today, I wanted to share my three favorite resources to learn NLP.

Is Learning NLP Still Worth It In The Age Of Generative AI? Here's why I think learning foundational NLP skills is a great investment.

If you're interested in building evaluations for generative AI applications, I highly recommend this blog post by Hamel Husain: Creating a LLM-as-a-Judge That Drives Business Results hamel.dev/blog/posts/llm… Here are a few reasons why:

One of my goals for 2025 is to write more. Today feels like a good day to start :)