PyCon AU 2025
2025-09-12..2025-09-16
Melbourne, Australia
PyCon AU ·
[PyCon AU 2010, PyCon AU 2015 OpenStack Miniconf, PyCon AU 2016, PyCon AU 2016 Science and Data Miniconf, PyCon AU 2017, PyCon AU 2019, PyCon AU 2023, PyCon AU 2024, PyCon AU 2025, PyCon AU 2026]
| title | author | length |
|---|---|---|
| Education Track Opening | Amanda J Hogan, Nicky Ringland, Alison Wong | 17m |
Announces practical conference details: a mask zone and dedicated seating, gendered bathrooms today with an all‑gender option on the ground floor, an AV office… | ||
| Data & AI Track Opening | Nic Crouch, Jack Skinner | 13m |
Opening sets logistics and scope: front-left of the ballroom is a mask zone, gendered bathrooms are in use today with a nearby gender-neutral option, and a… | ||
| Scientific Python Track Opening | Genevieve Buckley, Kai Striega, Charles Turner, Jo Basevi | 10m |
Scientific Python specialist track runs all day in Ballroom 2 and lays out concrete logistics and community rules: Pyon AU25 Wi‑Fi and attendee Discord for… | ||
| Life Beyond Pandas: Workflows with DuckDB, Daft, Polars, and Datafusion | Mimoune Djouallah | 26m |
DuckDB, Daft, Polars and DataFusion are compared by concrete outcomes: a workload shows 2× performance boost in under a year, a 150 GB CSV processes on a… | ||
| Why Teach the "Why"? | Jack Reichelt | 24m |
Describes concrete exercises and outcomes: students build a one-bit adder from truth tables through logic gates, scale it to a byte adder, and retain the… | ||
| Big Brains, Small Targets: Whole-Brain Image Analysis with Python | Ishrat Zaman | 25m |
An end to end Python pipeline registers whole mouse brains to an Allen atlas with brainreg, scales processing of ~40 TB (20 brains) via dask and HPC, and… | ||
| Going with the flow? Apache Airflow for operational-quality scientific workflows | Patrick Sunter, Michael Pegios, Daehyok Shin | 17m |
Uses Apache Airflow to turn existing Fortran/C and Python models into reproducible operational pipelines, integrating Conda environments, container images and… | ||
| The Duck and the DataFrame: A Data Engineer’s Journey with DuckDB | Ankur Jain | 27m |
Claims that DuckDB replaces heavy distributed tooling for moderate data sizes and shows concrete results: a dbt pipeline reads Hive‑partitioned Parquet on S3… | ||
| The Lab - Lessons from an Autism-Inclusive Learning Space | Maddie Mackey | 26m |
Weekly unstructured sessions run from a public library makerspace where teens use 3D printers, laser cutters, die cutters and library PCs to tinker, share… | ||
| A case study on building our first LLM feature – how to balance speed + quality | Vivek Katial | 26m |
A case study of building a production LLM feature in 1 month by a two‑person team: a pipeline that scores code‑review feedback for tone and specificity… | ||
| Modernizing Legacy: Wrapping a 25+ Year Computational Fluid Dynamics Codebase | Tishampati Dhar, Duy Nguyen | 29m |
A 25+ year-old C++ CFD model is wrapped with C-interop so Python notebooks can invoke compiled simulations and return temperature and algal‑bloom outputs… | ||
| Catching them all: teaching fundamental OOP concepts with Pokemon | Sujatha | 26m |
Claim: Pokémon maps directly to classes, instances, inheritance and polymorphism and shortens cognitive load by reusing a familiar domain. Uses Python and… | ||
| Capturing Flags with Microbits | Edwin Griffin | 26m |
Five weeks of learning delivering and assessing produces a classroom CTF that uses micro:bit radio networks, Python browser editors and USB serial to emulate… | ||
| On the Fly and On the Flight: Scientific Data Analysis Beyond the Beamline | Melanie Hampel | 30m |
Python suite actually produces immediate XDI outputs and reproducible pipelines, uses xarray for named multi‑dimensional arrays (handles up to 100 detectors… | ||
| Needle in the Haystack: Applied ML for Inspections of Power Infrastructure | Boaz Ash | 21m |
A full‑stack system is built to find rare faults across millions of aerial images captured at five frames per second from helicopter rigs over 58,000 km of… | ||
| Hierarchical Clustering: Finding the awkward reunions | Neha | 21m |
Claims a hierarchical random graphs generative model infers multilevel community structure and reproduces right-skewed degree distributions, high clustering… | ||
| From Matlab to Lambda: Transforming Structural Engineering Research Tools | Brooks Smith | 30m |
Porting CUFSM into pyCUFSM deployed to AWS Lambda yields a concrete, validated Python engine with a p90 solve time of 3.9 seconds for real workloads. The… | ||
| Who_Dunnit.xlsx – Teaching Python through Data Investigation | Renee Noble | 26m |
Uses Python in Excel or Jupyter to load spreadsheets into pandas, indexes the science row of a walk-time matrix, filters period‑3 timetables, merges attendance… | ||
| The Streamlit Experiment: building web dashboards with Yr 10-12 | Geoff Matheson | 23m |
Classroom trials use Streamlit with Github Codespaces and GitHub Classroom to teach Yr 10–12 Python; students build real dashboards, an image resizer that… | ||
| Time Series Analysis in Python: Easy Tools for Scientific Insight | Muhammad Sakib Khan Inan, Rubaiath E Ulfath | 26m |
Concrete demos show library APIs, data shape and results: Sktime follows a scikit‑learn style fit/predict workflow; Aeon sits on top of that ecosystem; Nixtla… | ||
| Taking Back the Streets: NaviLens and the Fight for Open Navigation | Stephen Tierney | 17m |
NaviLens ties low‑density DD tags (a 5x5 DD tag holds 24 bits — roughly 17 million IDs) to a central web service and those scans include location and device… | ||
| Data Structures: A learning journey | Izy Hogan | 30m |
Builds a skip list implemented skip list in Python from scratch, starting with a custom linked list and about ~70–80 methods across node, iterator/position and… | ||
| Object-Oriented Oncology: Making Sense of Complex Patient Journeys | Georgie Kennedy, Ghazaleh Niknam | 29m |
Claims a concrete implementation: a Python ORM that maps OMOP plus the oncology extension into rich CancerPatient/CancerDiagnosis objects and exposes episodes… | ||
| Beyond Vibes: Building Evals for Generative AI | Dilpreet Singh | 31m |
Claims an actionable evaluation workflow: a three-part approach of quick automatic metrics, sampled human preferences, and repeatable tests that run on every… | ||
| Building an electricity market model from scratch | Jack Simpson | 31m |
Builds an optimisation from a Victoria five‑minute snapshot and shows exactly what is modelled and why it matters: a two-generator toy grid with demand forced… | ||
| What’s in a Name? – Fuzzy Matching Techniques for Proper Nouns | Renee Noble | 27m |
Demonstration applies edit‑distance libraries, nickname banks and generative retrieval to real event data, comparing six libraries and finding rapidfuzz wins… | ||
| Takeaways from teaching Python in college | Vladimir Roudakov | 31m |
Over a year teaching college courses with Python, NodeJS, React, C, Drupal and WordPress yields concrete takeaways: require git for every subject and use… | ||
| PyEarthTools: Machine learning for Earth system science | Tennessee J Leeuwenburg | 30m |
An open source Python framework for building and running machine learning Earth system models, PyEarthTools supplies modular data pipelines (reprojection… | ||
| Scaling Security Anomaly Detection in Enterprise Knowledge Base with Dask | Isabelle De Backer | 28m |
The pipeline ingests metadata from SharePoint Confluence and Google Drive, canonicalises it to CSV, applies NMF/SVD and rule‑based plus sensitive‑pattern… | ||
| Student Showcase | Amanda J Hogan, Nicky Ringland, Alison Wong | 01h 41m |
(Amanda J Hogan, Nicky Ringland, Alison Wong) Showcase of student projects
https://pretalx.com/pycon-au-2025/talk/D7UHXY/
python, pycon, australia… | ||
| AI for Good: Using Responsible AI to Drive Social Impact and Inclusion | Dr.Swapnilsony Singh | 33m |
Claims a concrete Nexus framework that uses generative and agentic AI plus machine learning and human-in-loop reinforcement learning to close a measurable… | ||
| High altitude balloon imagery decoding in the browser with C, JS, and Python | Michaela Wheeler | 29m |
Browser builds of Horus and Wenet demodulators become real: WebHorus and WebWenet compile C demod code into Python modules, build those modules to WebAssembly… | ||
| Education Track Closing | Amanda J Hogan, Nicky Ringland, Alison Wong | 02m |
Announcements center on concrete logistics and perks rather than new ideas: a self-organized dinner system uses Discord and asks people to form groups of four… | ||
| Scientific Python Track Closing | Genevieve Buckley, Kai Striega, Charles Turner, Jo Basevi | 03m |
Announcement notes several protest rallies near Flender Street Station and Parliament House tomorrow that start at 11:00 a.m. with police likely to arrive… | ||
| Data & AI Track Closing | Nic Crouch, Jack Skinner | 10m |
The session delivers a wrap‑up of the track with sponsor shoutouts, thanks, logistics and invitations rather than technical substance, so decide on seeing it… | ||
| Conference Opening (Saturday) | Peter Hall | 18m |
Announcements cover practical conference details and policies: masks in front-left mask zones, a safety team reachable at 0431597233, Wi‑Fi and QR codes for… | ||
| Keynote: Ethics in AI: Building Fair and Just Intelligence Systems | Arwen Griffioen | 36m |
Shows concrete evidence and actionable methods: pulse oximeters overestimate oxygenation for Black patients leading to worse outcomes, a resume‑screener… | ||
| The Continued Developer Evolution | Derek Bingham | 29m |
Claims that AI agents speed up development but do not replace developers, backed by concrete demos: a Pac-Man+Snake game built from a single prompt with… | ||
| The Primordial Code | Nick Moore | 33m |
Cells behave like messy computers: DNA stores information in 3 billion base pairs per human, genomes fall from millions to a cost under $1000 to sequence, and… | ||
| An EV trip planner for Australia | David Colls | 29m |
An app called Tripler uses discrete optimization to minimize trip time while adding resilience constraints so plans survive inoperable chargers; it prototypes… | ||
| Scaling Python-Powered Machine Learning With Snowflake | Sheena Nasim | 29m |
Snowflake claims to run familiar pandas code at scale using Snowpark pandas, translating operations to SQL so 10 GB jobs run ~30× faster and a 1 TB job… | ||
| Beyond the Hype: Using AI Coding Agents Today | Michael Gribben | 30m |
CLI agents such as Claude Code and Gemini Code are the focus, with concrete habits: use searching the codebase and tool calls to read files, compress stack… | ||
| Building with Multiple Languages: A Python-First Approach | Ramon Perez | 33m |
Claim: keep Python as the orchestrator and offload hot paths to other languages. Shows concrete patterns and tools: Python as orchestrator, outback monitoring… | ||
| Escape the Python - Ten years of Australia's best Escape Rooms | Dr Tom | 31m |
A decade-old codebase runs across networks of Raspberry Pis using asyncio, migrated from Python 3.3/Debian 8 to 3.11/Debian 12 while the same core still powers… | ||
| Taking wheels mobile | Russell Keith-Magee | 29m |
Python 3.13 adds Tier 3 support for iOS and Android and the talk shows how to produce and test binary wheels for those platforms using CI buildwheel and a… | ||
| Args: Amazing or Approaching? | Evan Kohilas | 30m |
History and mechanics of Python arguments get traced to concrete claims and fixes: defaults behave oddly because default args created at definition time… | ||
| Guardrails: An Alternative View of Safely Working in Python | Christopher Neugebauer | 31m |
Argues that Python evolves guardrails—language features that constrain unsafe machinery and make correct use easier—by showing concrete replacements and tools… | ||
| My AI is slow. Make it faster! | Anthony Shaw | 31m |
Benchmarks show big differences: time to first token and total time matter more than topic, with some models returning in seconds while others spike up to a… | ||
| The Long Hello World | Noah Kantrowitz | 26m |
Follows print("hello world") all the way from the REPL through the tokenizer, PEG parser, bytecode compiler, and CPython VM into G nu C's write, the Linux… | ||
| Developer-friendly jobs across time zones when "out of hours" no longer exists | Suzannah Cooper | 29m |
Kraken Technologies ships over 100 versions per day to 25+ client environments in more than 10 countries and builds an internal Housekeeping Framework that… | ||
| Escape from Tutorial Hell | Sarah Reichelt | 26m |
Concrete checklist: pick platforms, frameworks and APIs then break the project into chunks (server, database, users/products/orders) and further into folders… | ||
| I learned everything about the match statement so you don't have to | Sam | 11m |
Shows concrete code: builds Circle, Rectangle and Square classes and implements an area method for shapes and a point-in-shape containment using positional… | ||
| Pipe-ifying Python: The how and why of interpreter hacking | Alex Mirrlees-Black | 28m |
Adds a concrete pipe operator to CPython by changing token and grammar files, rewriting the parse rule to be recursive, and lowering pipes into normal call… | ||
| What emergency service volunteering has taught me about incident management | Rachel Bunder | 32m |
Covers concrete practices and outcomes: the use of SAFER checks for personal readiness, the SMIAX template for Situation Mission Intent Execution… | ||
| Self-healing system for UI tests using ML | Andressa de Mello Cabistani | 32m |
A system is built around Python Playwright and pytest that wraps the page API with a self-healing locator which captures element fingerprints and, on timeout… | ||
| Lightning Talks (Saturday) | 01h 03m | |
A chaotic five‑minute set that actually delivers specifics: a craft demo of pipe‑cleaner earrings; common Python typin errors including using the wrong… | ||
| Conference Opening (Sunday) | Peter Hall | 19m |
Announcements list concrete logistics and claims so decide by specifics: live sprints run Monday–Tuesday and require a sprint ticket sprints Mon–Tue, there are… | ||
| Keynote: You are welcome in this world | Larene Le Gassick | 41m |
Keynote argues that inclusion is about people and tiny choices, not checklists, and proves it with concrete examples: a blind father using bump dots to operate… | ||
| MCP for dummies: Feed more context to LLM | Tarun Jain | 32m |
Claims that Model Context Protocol standardizes passing external context to models and reduces per-tool custom work and function-calling glue. Implementation… | ||
| The Death of Consequences | Christopher Neugebauer | 33m |
Claim: the early open‑source era ends because the community refuses to impose consequences. Evidence: permissive licenses plus platform convenience let… | ||
| Falsehoods Programmers Believe About Reality | Lilly Ryan | 29m |
Names, time, geography and categories repeatedly break systems: claim that more data = better model is false, and large language models are described… | ||
| Code Without Barriers: Inclusive Python Development with AI | Michelle Mei-Ling Sandford | 31m |
Shows concrete outcomes: the Code; Without Barriers launch trains 780,000 women skilled in year one with 373 job placements and a 500 million reach… | ||
| The Birdwatcher’s Guide to Optimised Tabular Data Pipelines | Jenya Bogacheva | 25m |
A real pipeline walk-through: a naive O(T·D) loop on 5 million detections and 300 species becomes a linear-ish sweep-line with sorting, dropping an XS case… | ||
| Found Font Family | Zain Afzal | 29m |
Builds a tiny font with cat sprites, binds them to kitty font mapped to private-use codepoints, and drives animation by a Python loop that cycles characters so… | ||
| Skill Issue | Benno Rice | 33m |
Claims that large language models generate code without understanding it and that they require billions with a B of investment per year to train are central… | ||
| Building Production-Ready MCP Servers: From Protocol to Patterns | Nitish | 26m |
Claims a practical path from Model Context Protocol to production: concrete Python stack using fast-mcp on FastAPI with Pydantic validation, demonstrated… | ||
| Tips and tricks: data science prototype into production | Ruth Luscombe | 32m |
Solves a multi‑dimensional knapsack with tens of thousands of items, up to a hundred user constraints, using Google OR‑Tools and deployed to AWS Lambda… | ||
| Sustaining Open Source Software: Tools for Change | Justin Warren | 29m |
Argues that maintenance beats creation and gives a checklist for assessing health of OSS projects with concrete fixes: add governance, conflict‑resolution, and… | ||
| A Chronicle of Digital Transformation : From Kitchen Chaos to Kubernetes | Dipendra Paudel | 30m |
Starts from paper, pens and 250 dispersed operations and builds a cloud‑native platform that actually ships: Python + Django backend, Docker and Kubernetes… | ||
| Unarchiving vulnerabilities and avoiding tar-pits | Caleb Brown | 29m |
Finds three vulnerabilities in Python's tarfile library including an arbitrary write path traversal that lets a crafted tar overwrite files anywhere it has… | ||
| Keeping Creativity the Focus in Visual Effects and Animation | Caitlin Wright | 26m |
Python is the actual glue: code that enforces filenames, directory structure, AOV exports and EXR delivery so animators, lighters and compositors just press… | ||
| Your datasets, under your control: Introducing the Mozilla Data Collective | Kathy Reid | 29m |
Claims that the web has hit a token crisis get quantified: an estimate of 114 trillion tokens used for a single large model versus 60–160 trillion web tokens… | ||
| Myths developers believe about open source security | Nicky Ringland, Tim Zhang | 29m |
Claim: a single dependency constraint can produce 10^81 possible graphs so your install is only the graph on this machine at this time. Evidence: ecosystem… | ||
| Pinchy the bestest boi | Hailey Bartlett | 27m |
Pinchy is a Raspberry Pi–powered six‑legged scorpion robot that claims autonomous mapping and navigation using SLAM and computer vision; the build uses… | ||
| Reinforcement Learning with Quantum Algorithms: Simulating Nematode Behaviour | Chris Zaharia | 31m |
Uses parameterized quantum circuits as a nematode “brain” trained by policy gradients with a parameter‑shift rule, running on a quantum circuit simulator and… | ||
| Reducing the "Oops Factor": Pipelines for a Secure Python Development Lifecycle | Simon Merrick | 29m |
Claims concrete fixes: a GitLab‑flavored CI/CD pipeline that adds secret detection, dependency scanning and SAST as gates and runs them in parallel; use git… | ||
| Lightning Talks (Sunday) | Nick Moore, Linda McIver, Ben Fitzhardinge, Elena Williams, Christopher Phillips, Morgan Barthol | 01h 05m |
Seven-minute demos report real outcomes: an Ultima 4 → Lua Anti/MineTest import that writes 16×16 block chunks and generates rounded rivers; a community launch… | ||
| Conference Closing (Sunday) | Peter Hall | 23m |
Closing remarks that list concrete outcomes and next steps: 584 registrations, 99 presentations (24 lightning, 9 student), 12 financial assistance recipients… | ||
| nothing matches | ||