Data Science podcasts

Zoe Weil discusses how recommendation and ranking systems can use in-session behavioral signals to model intent and optimize purchasing outcomes. The episode examines event-to-feature models, low-latency ranking infrastructure, evaluation against business outcomes, and the distinction between click relevance and useful recommendations. It also considers privacy tradeoffs in personalization and the relationship between advertising and organic revenue.

Hosted by Sam Charrington

This Nature Podcast explores how anonymized banking records can reveal patterns of financial abuse and inform support for victim-survivors. A second segment examines how opposing forces within Earth’s core can change the planet’s rotation and the length of a day. Research highlights cover Mercury’s contraction and cognitive resilience in predicting Alzheimer’s dementia.

INRIA researcher David Holzmüller explains how to choose among tabular foundation models, boosted trees and multilayer perceptrons. He discusses TabArena, TabICL and RealMLP, benchmark weaknesses, small-data performance, calibration and class imbalance. The conversation also examines text columns, explainability, time-series distinctions, synthetic pretraining and the trade-offs among computation, inference speed and predictive performance. Hosted by Ravid Shwartz Ziv and Allen Roush. Watch the full conversation on the publisher’s YouTube channel.

Hosted by Ravid Shwartz Ziv, Allen Roush

Ellen Zapata-Webborn discusses how mathematical research can help predict household energy use. The episode connects patterns in smart-meter data, human behaviour and the demands of electricity systems, illustrating how quantitative models can inform practical energy questions.

Tim Wahlberg of CAS speaks with Rafael Rosengarten about the role of curated chemical knowledge in AI-assisted drug discovery. They examine scientific data quality, harmonisation and expert curation, and consider how reliable information can help researchers formulate questions and evaluate answers from large language models.

Hosted by Rafael Rosengarten

University of Freiburg professor and Prior Labs co-founder Frank Hutter explains how TabPFN changed deep learning for tabular data. The episode examines in-context learning from synthetic pretraining, forecasting, architectural approaches to larger tables and the continuing roles of boosted trees and language models. Hutter also discusses moving research into a company, team autonomy, model licensing and building research organizations outside established technology hubs. Hosted by Ravid Shwartz Ziv and Allen Roush. Watch the full conversation on the publisher’s YouTube channel.

Hosted by Ravid Shwartz Ziv, Allen Roush

James Fergusson discusses artificial intelligence and data-intensive science, including how mathematical and computational approaches help researchers learn from complex datasets. The conversation connects research practice with the training needed to work across scientific disciplines.

Pleias co-founder Pierre-Carl Langlais describes language models built from documented open and public-domain sources alongside synthetic data. The conversation examines the SYNTH dataset, gaps in common web crawls, preservation of source material and ethical choices beyond copyright status. It also covers compact deployed models, benchmark incentives, reasoning-trace access and competing approaches to national AI development. Hosted by Ravid Shwartz Ziv and Allen Roush. Watch the full conversation on the publisher’s YouTube channel.

Hosted by Ravid Shwartz Ziv, Allen Roush

NVIDIA researcher Jean-Francois Puget discusses evaluating agent skills and distinguishing real improvements from benchmark overfitting. Drawing on Kaggle competitions, he emphasizes validation that separates development feedback from final evaluation. The episode also considers coding agents, multi-agent software development, model-release claims and his team’s approach to the ARC-AGI competition. Hosted by Ravid Shwartz Ziv and Allen Roush. Watch the full conversation on the publisher’s YouTube channel.

Hosted by Ravid Shwartz Ziv, Allen Roush

Roger Deane, Walter Silima and Sally McFarlane discuss the Inter-University Institute for Data Intensive Astronomy at its tenth anniversary. Topics include South Africa’s MeerKAT and SKA research capacity, the Ilifu cloud, CARTA, immersive data visualization and training. The episode offers a practical view of the people and infrastructure supporting African radio astronomy.

Jure Leskovec connects data-driven biological modelling with relational deep learning. Topics include multiscale virtual-cell models, treating databases as graphs and using in-context prediction across multiple tables. The discussion also covers benchmark evaluation, explanations, agent integration and the practical limitations of relational foundation models.

Hosted by Sam Charrington

Richard Gao discusses fitting spiking-network models to experimental data and dealing with different parameter combinations that explain similar observations.

Sara Solla discusses how low-dimensional descriptions can reveal structure in recordings from large populations of motor-cortex neurons.

Michael Hill explains how topological data analysis extracts structural information from large datasets. Ideas about connectedness, holes and shape, traditionally associated with pure mathematics, provide tools for recognising patterns as the scale of analysis changes. The discussion explores applications including breast-cancer research and questions about fair voting.

Heather Harrington explains topological data analysis and the construction of barcodes that summarize the shape of a dataset across scales. She describes how these tools reveal structure in complex information and how they can support studies of proteins and diseases such as cancer. The interview was recorded at the European Congress of Mathematics.

Carina Curto explains applications of topology to neural recordings and discusses mathematical models of recurrent networks.