PPMS - Submit Project Preferences

Workload Verification System

Many organisations, including UWA, use a workload model for fair and transparent allocation of work tasks to staff. Tasks accounted for in an academic workload model include unit coordination, project …

Many organisations, including UWA, use a workload model for fair and transparent allocation of work tasks to staff. Tasks accounted for in an academic workload model include unit coordination, project supervision and administrative tasks such as program chairs and head of department. A workload model comprises a set of tasks, each with a workload tariff (in, for example, hours per year); a set of staff each with available hours (based on FTE, contract type, leave etc); an annual allocation of staff to tasks. Most Universities currently use some form of (very large) spreadsheets to collect and review workload data, but this is hard to manage, validate and maintain. For example, school operations send each academic individual emails with screenshots of their workload allocation for 2026 and the academic emails a reply with any questions or corrections. This project is about responsible data management best practice, aiming to support transparency, protect privacy, reduce bias, and foster trust in workload data collection and reporting. The project seeks to support better processes for workload data validation by, for example: 1) Automated checks on the collected data to flag possible anomalies or errors, using a set of verification rules. 2) Human-in-the-loop checks: a process that allows individuals to review and provide feedback on data in the system for themselves or their staff. The system will input existing back-end data from spreadsheets and provide front-end interfaces and reports. For transparency, verification checks will be based on validation rules. A user-friendly review process will be developed to raise and check accuracy issues raised by individuals. Current PMC workload model documents and sample data will be provided.

Client: Daniela Roberts, PMC School Operations Coordinator

Location: UWA Crawley

Multiple Teams

IP: Creative Commons (open source)

Kayak Stroke ML Analysis

Forward thinking coaches increasingly rely on data-driven feedback to refine stroke mechanics. This project proposes building a cloud-native machine-learning solution leveraging AWS vision foundation models to detect and grade kayak …

AssetGuard AI: Intelligent Asset Usage & Compliance Portal

This project aims to build AssetGuard AI - a smart online portal that will help organisations quickly check whether planned equipment or machinery loads are safe for their infrastructure. Working …

An Online Platform for Intelligent Agent Mini-Contests

The project will build a website, potentially to assist the teaching of Intelligent Agent at CSSE. The platform needs to host student-developed game-playing agents and automatically runs the mini-contest between …

Book Borrow - Peer-to-Peer Book Lending Platform (Second Phase)

list books they are willing to lend or sell. Verified users can browse available titles, request book loans, or purchase books from other community members, promoting shared access to reading …

Social Tennis Setmaker

Looking for a build of software to generate social tennis sets for a local tennis club. Doubles social tennis (could also be utilised in Badminton, Padel and Pickleball) is a …

WMAA Charity Website Redevelopment Project

The WA Multicultural Assistance Alliance (WMAA) is a Western Australian not-for-profit organisation supporting multicultural communities, with a particular focus on crisis housing, family domestic violence support, public fundraising, and community …

The WA Multicultural Assistance Alliance (WMAA) is a Western Australian not-for-profit organisation supporting multicultural communities, with a particular focus on crisis housing, family domestic violence support, public fundraising, and community outreach. This project involves designing and building a modern, community-facing website that clearly communicates WMAA’s mission, services, and activities, while being easy for non-technical staff and volunteers to maintain. The website will act as a trusted public information platform for community members, donors, partners, and stakeholders. The core deliverable is a fully deployed, responsive website with a simple content management system (CMS) that allows authorised users to update key pages, publish news stories, and manage event photo galleries without coding. The site will include standard informational pages (Home, About Us, Services/Programs, News & Gallery, Contact/Donate), a contact enquiry form with anti-spam protection, basic analytics, newsletter subscription capability, and continuity with WMAA’s existing donation workflow. The solution must prioritise usability, low maintenance cost, basic security, and auditability, reflecting the operational constraints of a small charity without dedicated IT staff. The website would ideally be designed to support multilingual functionality, with English as the default language and the ability to support at least one additional language (for example, Chinese).

Client: Xiao Liang

Location: 10 Lake Street, Northbridge, Perth, WA

NDA Required

Multiple Teams

IP: Creative Commons (open source)

Language-Based Geographic Reasoning

This project aims to develop an AI-powered system that understands user queries and performs spatial analysis based on input map data. The system will use a large language model (LLM) …

Participation Marking App

An app for tutors to quickly and simply mark student tutorial/workshop participation. Participation marks are now a requirement for UWA units and this kind of app would free up limited …

H&S Incident Classification Model

This project involves developing a machine learning classification system for mining incidents using the damaging energies and critical risk frameworks. The system will be trained on approximately 1,120 pre-classified PLS …

This project involves developing a machine learning classification system for mining incidents using the damaging energies and critical risk frameworks. The system will be trained on approximately 1,120 pre-classified PLS incident records to automatically categorize incidents by energy type (human, gravitational, mechanical, electrical, chemical, thermal, etc.) and critical risk level (Fatal Risk, Permanent Impairment Risk, or Other). The model will include confidence scoring to flag uncertain classifications for human review, ensuring safety professionals maintain decision-making oversight. Deliverables include a multi-class prediction model, Python script supporting training and inference modes with CSV input/output, model retraining capability, and comprehensive documentation. PLS currently has over 7,000 unclassified incidents requiring approximately 350 hours of manual review. This tool will enable efficient, consistent analysis while directing safety expertise toward high-value risk assessment. Upon completion, PLS is open to releasing the model as public domain software (subject to IP approval) to benefit the broader mining industry and promote standardization of incident classification practices. Groups taking this project should have strong skills in machine learning, including natural language processing.

Client: Andrea Aurrecoechea Diaz

Location: West Perth

IP: Open to further discussion about best IP options.

Australian Virtual Computing Museum Interactive Exhibits

The Pearcey Foundation is developing a Virtual Museum of Australia’s Computing heritage with exhibits from around the country. The museum's goals are summarising, cataloguing and displaying the many historical computing …

The Pearcey Foundation is developing a Virtual Museum of Australia’s Computing heritage with exhibits from around the country. The museum's goals are summarising, cataloguing and displaying the many historical computing resources available in Australia through a virtual point of presence. The system would enable an enhanced virtual access to the detail of exhibits or to examine objects or original documents more closely than would be possible or permitted in a conventional museum. In addition, there is scope to include web-based or artifact-based (eg RPi) activities. WA has a rich history of computing with firsts in both hardware and software. See https://www.alex-reid.com/History/History-of-Computing-at-UWA.htm for some highlights. 2026 is the 50th anniversary of the Department of Computer Science at UWA, offering a great opportunity to share some of our computing history for future generations. There are many virtual museums internationally including the UK National Museum of Computing (https://www.tnmoc.org/) and Charles Babbage Institute at the University of Minnesota (https://cse.umn.edu/cbi). There are also many educational interactive activities for learning about computing eg https://www.csunplugged.org/en/ and https://www.llmsunplugged.org/ and https://uwaterloo.ca/computer-museum/blog/emulators Project: This project is open ended one - where students can propose what they will build, within the general scope of the Virtual Computing Museum. The group will identify a software or hardware application from the history of computing in WA and then design and prototype a virtual museum exhibit for it, including information and interactive activity(s). You are encouraged to consider a wide range of implementation options, leveraging recent advances in interactive web technologies, immersive visualisation/telepresence and AI or hands-on CS-unplugged activity maybe using IoT devices. The clients for this project are experts on UWA's computing history. Support will also be available from experts on HCI design, and contact with other computer museum groups around Australia.

Client: Terry Woodings and others

Location: Crawley

Multiple Teams

IP: Creative Commons (open source)

Mapping Global and National Political Networks

Seeing the Sangh (StS) is an online data collection project that has mapped thousands of organisations in the transnational HIndu far-right. This project had its initial release in December 2025 …

AI-Based Text Recognition: An End-to-End Challenge Using Noisy Historical Data

This project aims to enable linguistic analyses of historical texts that have been written in Fraktur font. It could be expanded and utilised for similar newspapers or even hand-written documents …

This project aims to enable linguistic analyses of historical texts that have been written in Fraktur font. It could be expanded and utilised for similar newspapers or even hand-written documents in the future to make digitised datasets available to non-German speaking researchers. The client, Oliver, is a PhD student at the School of Humanities. Part of his project is to critically analyse media content in a German newspaper from the 1930s. The paper was printed in Fraktur - font. Transkribus has been used to create a searchable text file from images uploaded to the program, however the results were not satisfying. The aim of this capstone project is to train existing AI - models for this font. The problems encountered so far are: - Accuracy of text recognition potentially affected by the quality of the scans due to age of the original content and general loss of resolution in the text images (TIFF images) - Recognition of text continuity due to the newspaper layout - Attempts to train existing models have been impaired by the above The project is to develop a prototype for an end-to-end ML pipeline for: - Identification of suitable AI models for my purposes - Development of an adequate methodology to train high accuracy of text recognition - Bulk uploading of images The project material includes challenging content that requires introduction to students in a critical and historically contextualised manner. Issues of AI safety, data security, confidentiality and broader ethical considerations will also be addressed and discussed throughout the project.

Client: Oliver Krumholz

Location: Crawley Campus

Multiple Teams

IP: Creative Commons (open source)

Fire Vulnerability App for Cultural Heritage Sites

Controlled and uncontroilled bushfires pose a significant threat to Australia's cultural heritage. This heritage extends back at least 65,000 years and continues today, as we produce the heritage of tomorrow. …

Controlled and uncontroilled bushfires pose a significant threat to Australia's cultural heritage. This heritage extends back at least 65,000 years and continues today, as we produce the heritage of tomorrow. Cultural heritage includes individual artefacts such as stone, wood, metal ceramic and other artefacts, rock art, burials, and the like. These individual artefacts are found at 'sites' where people in the past (and sometiems still today) lived, worked, made things, traded - all the things we do today. Collectively, these sites can make up a wider 'cultural landscape' that can hold 'intangible' heritage that is connected to artefacts, sites and even natural features. Aboriginal people have long used fire as a way to maintain Country and protect heritage. But today, bushfires are numerous, often uncontrolled, and seldom consider impact on cultural heritage sites. in order to provide Land Managers, Aboriginal Rangers, government departments (DFES, DBCA, DPLH especially), farmers and even private individuals with the tools to properly plan burns so that they not only cause no harm - but so they allow fire to be used as a tool for the positive management of heritage. One such tool would be an app (application) or similar that plots cultural heritage sites against vegetation, wind, landforms etc to show degrees of fire vulnerability for different landscape snad the heritage they contain. Public databases like DPLH's ACHIS and InHerit systems contain thousands of sites across Western Australia that can be used to trial such a system. I also have site information from the East Kimberley. A trial version could be done on 3-4 sample areas such as Perth Metro, Kimberley, parts of the South West of WA using a simple colour-cided stoplight system (red-orange-green) or similar to ai planning and as a preventative measure. This project could have one or both of a web-based expressions hsowing cultural heritage sites and fire vulnerability and/or an in-field app where land managers could add fire risk ratings when they visit specific cultural heritage sites or see any signs of recent fire and/or fire damage. This project will use publicly available data. During the project we will discuss issues of Indigenous Cultural and Intellectual property (ICIP) and Indigenous Data Sovereignty (IDaS) to foster awareness of the ethics of data usage.

Client: Sven Ouzman

Location: UWA Crawley campus

Multiple Teams

IP: Open to further discussion about best IP options.

AI Study planner

The main goal of this project is to develop an intelligent study planner that uses Generative AI (Gen AI) to automatically draft a personalised study plan based on student input. …

The main goal of this project is to develop an intelligent study planner that uses Generative AI (Gen AI) to automatically draft a personalised study plan based on student input. The AI will take data like course units, prerequisites, availability, and number of semesters to create a draft plan. Students can then modify the plan, and the system will re-evaluate it, identify errors, and suggest improvements. The study plan will be exportable, and changes will be saved throughout the process. The primary focus of this project is on building the study planning engine/model, with the interface and other aspects being secondary priorities. Key objectives include: AI-Generated Study Plan: Automate the creation of a personalised study plan based on student input. Student Customisation: Allow students to modify and adjust the AI-generated plan. Re-evaluation & Error Detection: Re-evaluate the study plan and highlight potential issues (e.g., scheduling conflicts, unmet prerequisites). Save & Export: Enable saving the study plan at every step and exporting it to Excel/PDF. Project Tasks Project requirements gathering and initial designing, Defining the AI model and technologies to be used and defining the scope of input data (courses, units, availability, semesters, prerequisites) and output data (study plan, errors). AI Model Training & Integration Development of other core functionalities Provide a visual representation of the study plan. Design a modification interface to let students adjust the plan and see real-time feedback from the AI.

Client: Maryam Vettoor

Location: You can also contact [email protected] or [email protected]

IP: Creative Commons (open source)

Web-based 3D visualiser for scientific data

<a href="https://mxrap.com/">mXrap</a> is a geotechnical analysis software platform in which suites of apps can be developed. It is used primarily in underground hard metal mines by geotechnical engineers for tasks …

mXrap is a geotechnical analysis software platform in which suites of apps can be developed. It is used primarily in underground hard metal mines by geotechnical engineers for tasks such as analysing and interpreting mining-induced seismicity. It would be useful to have the ability to export information from mXrap's 3D visualisation to a web-based system, so that key points of analyses or reports could be easily viewed by other people who do not have access to mXrap itself. This project involves the design and creation of a web-based system for 3D visualisation of exported data. In the simplest form, this data would represent a static scene made up of multiple data series. Each series has a certain type of representation in the 3D view: surfaces (polygonal data), lines, points, or texts. Each series also contains associated scalar data which is represented by colouring the 3D representation, for example the vertices of a surface may be coloured to show a calculated stress value at that point on the surface. The mXrap program uses the VTK library for visualisation, so it may be interesting to consider VTK.js when planning your implementation of this system, but it is not a requirement. Rendering large numbers of points represented by sprites is an important feature that is not covered by the core VTK library, in mXrap some scenes involve many millions of points.

Client: Matthew Heinsen Egan

Location: Australian Centre for Geomechanics, Civil and Mechanical Engineering, Room 273

Multiple Teams

IP: Open to further discussion about best IP options.

A web-based system for inspecting ground support testing data

Ground support is used in mining and civil engineering to stabilise rock masses, crucial for preventing rockfalls and to minimise the risk of injury in underground mines or tunnels. Ground …

Ground support is used in mining and civil engineering to stabilise rock masses, crucial for preventing rockfalls and to minimise the risk of injury in underground mines or tunnels. Ground support designs use elements such as rock bolts, mesh, and shotcrete. For each of these elements there are different products available, with different performance characteristics. Engineers must decide which products to use for their specific projects in order to meet their safety and budget requirements. Most ground support suppliers have comprehensive databases of their ground support elements' performance. This information is shared regularly with their clients involved in ground support design activities. For engineers engaged in ground support design it would be more efficient and useful to have a centralised repository that combined data from many ground support suppliers. This would enable users to easily compare products from different suppliers, and to search for all products which met their particular performance requirements. This project will consist of the design and implementation of a database for storing ground support element test data, and a web-based system for accessing that data. Ideally the web-based system would include the ability to search elements based on particular performance characteristics, and be able to show graphs of the testing data for particular elements.

Client: Matthew Heinsen Egan

Location: Australian Centre for Geomechanics, Civil and Mechanical Engineering, Room 273

Multiple Teams

IP: Open to further discussion about best IP options.

Machine Learning for Predictive Maintenance

This is a project suitable for students with experience in Data Science and training Machine Learning (Deep Learning) models. This project includes the possibility (no promises) of working on a …

This is a project suitable for students with experience in Data Science and training Machine Learning (Deep Learning) models. This project includes the possibility (no promises) of working on a conference paper if good progress is made during semester 1 (https://irai2026.ieee-ies.org) This project has industrial interest; we will be working alongside Programmed (https://programmed.com.au/service/facility-management/) to better understand the challenges of implementing machine learning based predictive maintenance. What the project is: We would like to train a model for anomaly detection in data coming from a centrifugal pump, with the hypothesis that clustering or increased frequency of anomalies is indicative of imminent pump failure. If the hypothesis is born out then such a model could be used, with real-time data, as an early warning that a duty pump was approaching failure, giving time to switch over to a stand-by pump and increasing overall facility reliability. It is anticipated that getting sufficient training data to build a functional model may be a challenge; therefore a secondary hypothesis is that a model built using transfer learning (trained initially on a wider pool of equipment data and then with a top-layer retrained on pump data) will be better than a model trained only on pump data. Deliverables would be: well documented python code, a short theoretical explanation of the learning methods used and a "user manual" such that a person familiar with python but not part of the programming team, could either retrain the model with additional data sets or test the model with a data set that was not part of the training data.

Client: Pete Whittaker

Location: UWA, School of Engineering, Department of Mechanical Engineering

Multiple Teams

IP: Creative Commons (open source)

View Projects or Submit Project Preferences

CITS5206 - Information Technology Capstone Project

Group Information

Only use this form for a team's preferences

Your Selected Projects

Only use this form for a team's preferences

How to Rank Projects:

Available Projects

Workload Verification System

Kayak Stroke ML Analysis

AssetGuard AI: Intelligent Asset Usage & Compliance Portal

An Online Platform for Intelligent Agent Mini-Contests

Book Borrow - Peer-to-Peer Book Lending Platform (Second Phase)

Social Tennis Setmaker

WMAA Charity Website Redevelopment Project

Language-Based Geographic Reasoning

Participation Marking App

H&S Incident Classification Model

Australian Virtual Computing Museum Interactive Exhibits

Mapping Global and National Political Networks

AI-Based Text Recognition: An End-to-End Challenge Using Noisy Historical Data

Fire Vulnerability App for Cultural Heritage Sites

AI Study planner

Web-based 3D visualiser for scientific data

A web-based system for inspecting ground support testing data

Machine Learning for Predictive Maintenance

Ranking Guidelines

Consider Team Skills

Evaluate Requirements

Rank Strategically