Mr Peter Marendy1
1Qcif Ltd, St Lucia, Australia
Biography:
Peter is currently employed as Head of Data and Software Solutions at the Queensland Cyber Infrastructure Foundation (QCIF), which provides eResearch infrastructure and services for Queensland research institutions and contributes to the National Research Infrastructure.
In this role, Peter is responsible for leading a team that delivers innovative, high quality, and time driven results for a wide variety of research programs from within universities, research institutes, and commercial companies. The team has expertise in workflows, specialised computing, data capture and management, and working with sensitive data.
Prior to his role with QCIF, Peter led the Microsystems research within CSIRO’s Cybernetics research Group. During his time at CSIRO, Peter also led projects such as Bees with Backpacks, Optimising Pollination, Smart Hives, Probing Biosystems – Implantables, Brain Implants, and Smart Helmets for optimising the timing of Cranioplasty.
Peter is also a member of the RSE-AUNZ steering committee which aims to build awareness of the diverse Research Software Engineer (RSE) roles, to connect volunteers in the RSE domain, and to help build practical solutions for the RSE Community.
Peter brings his experience in team/capability management, project management, and customer focused collaboration and relationships. A people focused leader with broad knowledge across multiple domains with a focus on cutting edge development and applications. An excellent technology communicator to the public, academia, government, grant bodies, and industrial stakeholders.
He also has more than 20 years of software engineering experience across multiple domains, including Digital Agriculture, Energy, Food and Nutritional Sciences, Health, Marine Sensing, Robotics, and Visual Analytics, in research and innovation environments.
Abstract:
As data privacy regulations tighten and datasets grow increasingly siloed across institutions, Federated Learning (FL) offers a transformative approach to collaborative machine learning in the eResearch domain. This talk introduces the core principles of Federated Learning—how it enables model training across decentralised data sources without moving the data itself—and explores its relevance to research environments where data sensitivity, ownership, and locality are paramount.
We will outline the main types of FL, including horizontal, vertical, and federated transfer learning, and discuss their applicability to real-world research scenarios. Drawing from our own experience, we’ll share insights into evaluating and selecting FL frameworks, the practical challenges of setting up federated infrastructure, and lessons learned from early implementations.
Finally, we’ll look ahead to where the field is going: the growing role of privacy-preserving technologies like differential privacy and secure multiparty computation, the need for standardisation, and the potential for FL to unlock new forms of cross-institutional collaboration in science and academia.
Whether you're a researcher, data scientist, or infrastructure specialist, this session will provide a practical and forward-looking perspective on how Federated Learning can reshape data-driven research.