Hi everyone,
I have some exciting news I wanted to share with the Hugging Face community, as this is where my journey really started.
I was just accepted to speak at PyData Global 2025 on my talk, “I Built a Transformer from Scratch So You Don’t Have To”! 
I was thinking back, and it all started with the Transformer from Scratch tutorial repo that I first uploaded here on the Hub. That was the exact project I submitted with my proposal, and it’s so rewarding to see that commitment to first-principles understanding resonate with the wider community.
The talk will be a deep dive into demystifying the attention mechanism and the core components of the transformer, making it accessible for everyone.
For those interested in going even deeper, this work has since evolved into my more recent repos implementing advanced architectures like DeepSeek’s MLA and MoE, which I see as a natural extension of the talk.
I’m so grateful for the open-source culture here that makes this kind of self-directed research possible. Hope to see some of you at the virtual event!
Jen