Talks
Below you can see a list of prior talks (chronological order), as well as upcoming talks, and their resources. You can click here for a list of all suggested talks.
We try to have a talk every other Thursday at 12pm.
Upcoming
Titles are placeholder.
- 2026-04-02* | "Open code from a journal submissions perspective" by Erik van Sebille ( Discussion)
- Abstract: Many academic journals require research code to be available when you submit a manuscript to them. But how do you organise that? Where do you put the code in such a way that you can still make changes to it, and that it is also archived for the future? In this talk, I will show how you can combine GitHub with a Zenodo workflow to create a DOI on every Release of your code. We will also have time for a more general discussion on the opportunities and difficulties of publishing code with manuscripts.
- 2026-04-16* | "Data processing with CDO" by Oliver Mehling ( Discussion)
- 2026-04-30* | TBD
- 2026-05-14 | None (Public holiday)
- 2026-05-28* | TBD
- 2026-06-11* | TBD
- 2026-06-25* | TBD
*Unconfirmed location (might be back to 6.07)
Prior
2025-02-25 | "Git and GitHub: A time machine for every research project" by Nick Hodgskin
Recording, Slideshow, Discussion
Abstract: Managing versions of a document is easy - just copy, paste, rename (often appending _old or _draft1 to the end), and you're good to go. But what happens when your research involves dozens of inter-related scripts, datasets, and perhaps even collaborators? Suddenly the "copy paste" version management becomes a nightmare. How can we keep track of changes, prevent lost work, and confidently explore new ideas without fear of breaking everything? Enter Git, a version control system that acts like a time machine for your research. In this session, we’ll demystify Git, explaining its core concepts - commits, branches, and repositories - and show how it can be used by independent researchers as well as large teams. We’ll also introduce GitHub - a powerful platform for sharing, reviewing, and archiving research code. The talk is based off the highly polished "Version Control with Git" course from Software Carpentry.
2025-03-06 | "Data Science Template: Go from zero to hero with a fully initialised template research project (with sensible defaults) in a minute" by Franka Jesse
Recording, Slideshow, CCDS Documentation, Showcase Example, Discussion
Abstract: We've all been there – eager to start a new research project, only to end up with a chaotic mess of notebooks like 'final_data_analysis_REALLY_FINAL_version6.ipynb' and data files scattered across our desktop like digital confetti. Fear not! In this lunch talk, I will introduce the "Cookie Cutter Data Science" template (CCDS for short) from Data Driven. It's a quick and painless way to set up your research project in under a minute—with a logical folder structure, sensible defaults, and zero regrets. We will also cover how to integrate it with Git for better version control (and fewer "final" versions). Whether you're handling climate model outputs, wrestling with observational data, or training a machine learning model, this template will save you from future headaches - and impress your collaborators with your beautifully organized project.
2025-03-27 | "Python Tidbits: Small Python tips, tricks, and packages you wish you knew about yesterday" by Nick Hodgskin
Recording, Notebook, Notebook Download, Discussion
Abstract: Sometimes the most impactful, timesaving features of a programming language are small tips, tricks, and patterns that you pick up along the way. When you finally come across them, you're left thinking "if only I knew about this earlier!". In this talk, we'll go through many of these lesser-known Python tricks - most of which are built directly into the Python language or standard library – to supercharge your Python skills. This talk will be fast paced covering a wide range of features, focusing on "tidbits" that can be explained in a few minutes each.
2025-04-17 | "Conda Environments: A tool for reproducible, collaborative, and portable research" by Nick Hodgskin
Recording, Slideshow, Discussion
Abstract: Every research project has different requirements. In software this comes in the form of different software dependencies. As we get involved with different research projects, we must manage these different (perhaps conflicting) dependencies - which can be an absolute nightmare if you don't have the right tools! In this talk we go over Conda environments which allow us to isolate dependencies so that we can work in isolated sandboxes for each of our projects. We can easily share these environments, whether it be with collaborators, with readers of your article, or if you want the same environment on your machine and on HPC. Conda environments are one of the key parts of reproducible research. At the end of this talk we also take a brief look at Pixi, a modern alternative to Conda environments which is picking up steam.
2025-05-15 | "VS Code: Your All-in-One Scientific Coding Environment" by Willem Kroese
Recording, Slideshow, Discussion
Abstract: As scientists, our projects span multiple languages, diverse config/data files, notebooks, local scripting and cloud or cluster work. In this hands-on workshop you’ll discover how Visual Studio Code brings all these workflows under one roof. We’ll kick off with a quick tour of VS Code’s core features (command palette, terminal, workspace management), then dive into essential settings and extensions. You’ll pick up time-saving tips & shortcuts (multi-cursor edits, snippets, Zen Mode), get a peek at remote development over SSH, and watch GitHub Copilot assist your coding with suggestions. Finally, as a real-world bonus, I’ll show you how I turned VS Code into my local LaTeX editor. By the end, you’ll have a personalized toolkit of settings, extensions, and workflows to supercharge your VS Code experience. Bring your laptop if you want to follow along!
2025-06-12 | "A Roadmap to the Pangeo Ecosystem" by Nick Hodgskin
Recording, Slideshow, Discussion
Abstract: We've all heard the name Pangeo - at a conference, or as a mention from the documentation of one of our tools - but what is it? Why are many climate scientists excited about it? It this talk we explore the Pangeo community and the ecosystem of packages it provides, going over its core mission and why it's needed in the next era of climate science where working with "big data" is becoming the norm. In this talk will give you a roadmap of the different packages and resources available in the Pangeo ecosystem - so that you can identify useful tools to investigate for your workflows and use-cases.
2025-09-04 | "Technical Difficulties Round Table"
Abstract: In our scientific workflows, we all encounter technical difficulties. How can I easily transfer data from a server to my computer? How can I easily iterate using a notebook on an HPC? How can I share data in a cross-institution collaboration? We'll use this session as a round table (or maybe "support group") where we can share our frustrations. Let's share our workflows, and voice the difficulties that we encounter so that we can learn from each other. This session will not be recorded so that people can share openly - an anonymised summary will be uploaded after discussion.
2025-09-18 | "Programming in Julia: Tools for Dynamical Systems and Climate" by Reyk Börner
Recording, Slideshow, GitHub Repo, Discussion
Abstract: As a climate scientist, chances are you’re working with some model written in Fortran and analysing data using Python. While Fortran is fast but a bit clunky, Python’s simplicity can come at the cost of running slowly. Julia, an increasingly popular programming language, seeks to offer the best of both worlds by combining speed with an intuitive, Python-like syntax. In this seminar, I’ll share my experience with co-developing my first Julia package (CriticalTransitions.jl) as someone who used to write messy Python scripts. We’ll explore the basic use, pros and cons, and what functionality is out there for dynamical systems analysis and climate modelling.
2025-11-13 | "Taming the Scheduler: Primer and Good Practices for Slurm on HPC Systems" by Lucas Esclapez
Repo with Jupyter Notebook, Recording, Slideshow, Discussion
Abstract: State-of-the-art climate research increasingly depends on high-performance computing (HPC) platforms to process large datasets and to run simulations of ever more complex models. Unlike personal computers, HPC systems provide a shared pool of computational resources accessed by many users simultaneously. Efficient and fair use of these resources is managed by a scheduler — a key component that can be both powerful and, for new users, somewhat opaque or frustrating. In this talk, I will introduce Slurm, one of the most widely used workload managers in the HPC community. I will explain how Slurm allocates resources, manages jobs, and maintains system efficiency. The session will also cover good practices for submitting, monitoring jobs and optimizing resource usage.
2025-01-22 | "Xarray on HPC and in the cloud: Working with CMIP6 data" by Tim Herman
Recording, Slideshow, Discussion
Abstract: The simulations of global climate models participating in CMIP6 provide a wealth of information but are challenging to analyse due to the wide variety of models and conventions and the large volume of data, especially when looking at higher resolution simulations or multiple simulation variants per model. Traditional workflows involve locally downloading CMIP6 data from ESGF servers or analysing data directly at those servers. With the introduction of CMIP6 data on Google Cloud, faster and more transparent cloud-based analyses are now also possible. Packages like xmip help analysing the data across models with different conventions in a structured way. In the In this talk I will explain the advantages of using cloud-based CMIP6 data and run you through a basic but neat demo workflow, based on examples available in the Pangeo Gallery and my own experiences.
2026-02-05 | "Xarray Series Part 1: Fundamentals of Xarray" by Nick Hodgskin
Recording, Slideshow, Discussion
Abstract: In the first instalment of this series, we follow the amazing Pythia Foundations online resource to learn the fundamentals of Xarray in the context of the geosciences. By the end of this talk, you will know the basic data structures in Xarray, be able to inspect DataArray and Dataset objects, as well as read and write netCDF files. Most importantly - this talk will start to show how Xarray based workflows can greatly simplify analysis and data processing of multidimensional climate data. There is no required pre-requisite knowledge to attend this talk. Whether you're an experienced Python user, or new to Python - everyone is welcome.
2026-02-19 | "Xarray Series Part 2: Computational Patterns with Xarray" by Nick Hodgskin
Recording, Fixed "Computations and Masks with Xarray" notebook, Original "Computations and Masks with Xarray" notebook, Discussion
Abstract: Having a grasp on the tools that Xarray gives you is the difference between a performant one-liner, and nested for-loops that take minutes to compute. In this talk, we go through the computational patterns in the Xarray toolbox, breaking them down and showing when to use them. These powerful abstractions from array programming allow for the efficient working with array data, with many concepts also extending to Numpy and other array libraries. We will be continuing with the amazing Pythia Foundations resource for this talk, as well as looking at the Xarray documentation.
2026-03-05 | "Supercharge your Python environments using Pixi" by Barber Vos
Recording, Pixi Documentation, Slideshow, Notes, Discussion
Abstract: Many research projects depend on a specific computational environment. Over time, changes in dependencies, Python versions, or operating systems can make scripts harder to run or share, especially when working across different machines or with collaborators. In this talk, I will introduce Pixi, a drop-in replacement for Conda which provides much better workflows for managing project environments and dependencies. Whether you work alone or in a team, Pixi helps you see exactly what your project depends on, collaborate without environment headaches, and reproduce other researchers’ work.