I occasionally give talks and interviews. This page is an attempt to collect the publicly available media in one place. Feel free to let me know if there are any that I’ve missed.
Those that are available on YouTube are also collected in a playlist.
Slideslive has an overview page of all the content I’ve contributed to conferences where they were in charge of recording and streaming, including ICLR, ICML and NeurIPS.
Interviews and podcasts
First Principles (July 2026)
An interview hosted by my friend and former colleague Marta Garnelo. There is an accompanying blog post on the Fundamental website.
History of Diffusion (May 2025)
An interview hosted by Slater Stich of Bain Capital Ventures, part of a series of three interviews on the history of diffusion models. The other interviews are with Jascha Sohl-Dickstein and Yang Song, who actually invented this stuff (all I do is blog about it), so I recommend checking them out as well.
NeurIPS chat: diffusion models for language (December 2022)
A brief chat with the folks from Zeta Alpha about diffusion models for language. This was shortly after we put the CDCD paper on arXiv.
Talks and lectures
Diffusion models for image and video generation
I’ve given a few talks providing an overview of what goes into training diffusion models for image and video generation at scale. These differ in length but cover the same things.
At Off The Radar, a tech conference in Brussels, Belgium (June 2026) – not sure why they thought an AI-generated thumbnail was a good idea though 🤭 this is the 20-minute version:
At AI Engineer London (April 2026), 30-minute version:
At ML in PL in Warsaw, Poland (October 2025), 50-minute version:
Summer school lectures on diffusion models
I’ve given several lectures on diffusion models at summer schools. These are very similar to each other (with some updates in the more recent editions). Like most of my materials, the focus is squarely on building intuition, but these lectures skew a bit more academic.
At The Eastern European Machine Learning Summer School in Cetinje, Montenegro (July 2026), note that this is a video covering the entire first day of the summer school, my lecture starts at approximately 6:14:00 into the video:
At The Mediterranean Machine Learning Summer School in Split, Croatia (September 2025):
At The Eastern European Machine Learning Summer School in Novi Sad, Serbia (July 2024):
I also gave an abridged version of this talk at the ICML 2024 workshop on controllable video generation (July 2024), which you can find on SlidesLive.
Multimodal iterative refinement
At the NeurIPS 2024 workshop on adaptive foundation models, I gave a talk about multimodal iterative refinement: some thoughts about how we deal with different modalities in different ways. This can be found on SlidesLive.
Summer school lecture: how to train neural networks effectively
For the 2023 edition of the Mediterranean Machine Learning Summer School in Thessaloniki, Greece (August 2023), I was asked to give a talk about the practice of training neural networks. It is a collection of tips, tricks and insights that I’ve amassed over the years, interspersed with examples from my own work and others’:
DeepMind x UCL deep learning lecture: convolutional neural networks
For the 2020 edition of DeepMind and UCL’s deep learning lecture series, I gave a lecture on convolutional neural networks for image recognition (June 2020).
Generating music in the raw audio domain
At the London Machine Learning Meetup, I gave a talk about generating music in the raw audio domain (June 2019). I later used this as a basis for a tutorial at ISMIR 2019, and a follow-up blog post.
Two alternative versions of this talk are also available on SlidesLive, recorded at ECML-PKDD 2020, and at ICLR 2021.
Classifying galaxies and plankton with deep neural networks
A talk about my work on computer vision problems in astronomy/biology in the context of Kaggle competitions, at the British Machine Vision Association workshop (February 2018).