Back-to-Backprop - Neural Networks From Scratch

---

Start: 28.05.2025 @ 18:00

---

We're excited to welcome Jovan Krajevski as a speaker at this month’s PyData Skopje meetup!

🧠 Talk Title: Back-to-Backprop: Neural Networks from Scratch

In this session, Jovan will revisit the foundations of deep learning. For attendees new to the topic, he will provide an intuitive introduction to what neural networks are and how they learn. For those with experience, the talk dives deep into the inner workings of the backpropagation algorithm — the engine that powers learning in neural networks.

To wrap up, Jovan will live-code a basic computational graph, inspired by PyTorch and TensorFlow, aiming to demystify the "magic" behind deep learning frameworks.

👨‍💻 About the Speaker: Jovan Krajevski is a computer engineering graduate with years of professional experience in the field of machine learning. His work has been particularly focused on applying deep learning to computer vision. He is currently pursuing a master’s degree at FINKI, where he is exploring the field of Bayesian statistics.

Don’t miss this hands-on, from-the-ground-up look at how neural networks really work!