Computer Science
Applications for 2025-2026 open on 1 July 2025
Understanding how machines learn to sample protein shapes and function
Project code: SCI028
Supervisor(s):
Discipline(s):
School of Chemical Sciences
School of Computer Science
Project
This project will investigate how machine learning (ML) models learn to predict and simulate protein structures and infer their function from structural data.
The role
This project introduces students to key concepts in ML/deep learning, 3D data representation, and protein structure-function relationships. You will work with tools for statistical data analysis, and ML learning frameworks for clustering and classification of protein molecules, alongside protein sequence and structure datasets, to explore how algorithms:
i) "See" and classify molecular features for drug design
ii) Assert molecular functions and dysfunctions in disease
Ideal student
This project is ideal for students with interests in AI, bioinformatics, or structural biology who want to bridge the gap between computational models and biological insights.
This project is in collaboration with academics in Denmark specialised in smFRET spectroscopy, opening up opportunities to network.
Pacman Assembly Arena: A Web-Based Assembly Programming and Debugging Tool
Project code: SCI053
Supervisor(s):
Discipline(s):
Ngā Motu Whakahī
School of Computer Science
Project
This is a Ngā Motu Whakahī scholarship specifically targeted toward students who whakapapa Māori or are of Pacific heritage.
The role
Help design and build Pacman Assembly Arena, a fun web-based platform that teaches assembly programming through interactive gameplay. You’ll develop features like a visual debugger, step-by-step code execution, and a game engine where student-written assembly code controls a Pacman or ghost.
This project blends web development, game logic, and low-level programming in a creative, educational tool. You'll be supported by a Māori or Pacific supervision team and take part in regular hui throughout the summer.
Enhancing Student Engagement and Rewarding Mechanisms via Web 3
Project code: SCI054
Supervisor(s):
Discipline(s):
Ngā Motu Whakahī
School of Computer Science
Project
This is a Ngā Motu Whakahī scholarship specifically targeted toward students who whakapapa Māori or are of Pacific heritage.
Web3 blockchain technology offers new ways to increase student engagement in university lectures. By using smart contracts, universities can create token-based reward systems where students earn tokens for attending and participating. These tokens can be exchanged for benefits like extra office hours or supermarket coupons.
The role
Your task in this project is to build a simple, user-friendly wallet application that enables students to manage and redeem their tokens easily on our blockchain platform.
The historical and living memory of Ōkahumatamomoe and the Ngāti Whātua Ōrākei Eviction through VR, Indigenous Storytelling and Game Design
Project code: SCI055
Supervisor(s):
Discipline(s):
Ngā Motu Whakahī
School of Computer Science
Project
This is a Ngā Motu Whakahī scholarship specifically targeted toward students who whakapapa Māori or are of Pacific heritage.
The role
This summer research project will explore and develop an immersive interactive experience using VR, Indigenous storytelling, and game design. This project seeks to uncover, restore, and activate mātauranga associated with Ōkahumatamomoe and Ngāti Whātua Ōrākei. It will also strengthen identity, address historical trauma, and ensure future generations inherit a truthful and empowered narrative.
The researcher will be supported in this project by two academic supervisors from Waipapa Taumata Rau and a mentor from Ngāti Whātua Ōrākei.
Beyond Prediction: Data Science summer projects
Project code: SCI057
Supervisor(s):
Assoc Prof Lara Greaves (Ngāpuhi)
Prof Mark Gahegan (Pākehā)
Eric Marshall (Ngāpuhi)
Tori Diamond (Ngāpuhi)
Assoc Prof Phil Wilcox (Ngāti Kahungunu, Rongomaiwahine, Ngāti Rakaipaaka
Otago University, Assoc Prof Andrew Sporle (Ngāti Apa, Rangitāne, Te Rarawa)
Discipline(s):
Ngā Motu Whakahī
School of Computer Science
Statistics
Project
Up to five students
This scholarship will be hosted by Ngā Motu Whakahī and is specifically targeted toward students who whakapapa Māori.
We are offering up to five summer projects for tauira Māori, through Beyond Prediction, an MBIE-funded data science platform.
The role
Students will be matched with a supervisor and topic based on their skills and interests. Past students have explored population data, epidemiology and health data, statistics and maths education, Māori data sovereignty, AI, data visualisation, and applying tikanga and mātaraunga to data science.
Ideal student
We are particularly interested in students with coding skills. Please get in touch with Lara to discuss topics/supervision.
Applied adiabatic quantum computing
Project code: SCI166
Supervisor(s):
Calude
Delmas
Discipline(s):
School of Computer Science
Project
Several real-world applications in optimization now need adiabatic quantum computing algorithms.
The role
Our team needs mathematical computer science students to help develop QUBO formulations. Two examples of problems are:
(a) Cost savings of flying drones in asymmetrical networks with repeated destinations
(b) Circuit/network embeddings that minimize area or cumulative wire lengths.
Requirements
- Python coding
- Reporting (latex preferred)
- Report and code handed at the end of the project
- Attend regular meetings
Interactive display with acoustically trapped particles
Project code: SCI167
Supervisor(s):
Discipline(s):
School of Computer Science
Project
Imagine drawing animated pictures in thin air. With a grid of tiny ultrasound speakers we can make a pin-sized bead hover and zip around so quickly that brief flashes of light paint a little 3-D shape you can see from every side.
The role
In this project you will build this levitation setup to allow 3-D animation with a particle floating mid-air, then take it further by adding simple cameras, microphones or gesture sensors so the floating image bends, spins or morphs as soon as someone waves, points or speaks.
The research challenge is to tie the user’s actions to the particle’s flight path in real time, turning a neat physics trick into a responsive mid-air display.
Check out a demo of the platform here:
https://www.instructables.com/Acoustic-Levitator/
Requirements
Interest and experience with electronics design (PCB design and soldering), FPGA programming and embedded-system coding on RPI or other microcontrollers
Investigating and predicting the severity of respiratory infections using AI
Project code: SCI168
Supervisor(s):
Discipline(s):
School of Computer Science
Project
Respiratory illnesses pose a recurring challenge to healthcare systems each year. Severe infections often require hospitalisation and can result in patient death. During the winter season, various respiratory viruses circulate, including influenza, which is responsible for approximately 500 deaths annually in New Zealand.
Hence, the severity of these infections can be measured by their risk for individuals to have a severe disease progression, but also based on their impact on the healthcare system, which is prone to collapse, especially during the winter season.
The role
Building on decades of research on respiratory infections in New Zealand, the Triple R project aims to investigate the severity of influenza and other respiratory infections. Its goal is to develop models to better inform patients and communities at risk and predict severe disease outbreaks in order to support proactive healthcare management.
The candidate will analyse multiple datasets from the collaborative Triple R project and assess their suitability for integration into individual-level risk models or disease outbreak forecasting.
Requirements
A main question of this student project is to what extent AI can be used to support these efforts. Therefore, the candidate should have experience in training and evaluating supervised machine learning models. Proficiency in Python is preferred, alternatively, experience with R would be a minimum requirement.
Adaptive AI for Augmenting Collaborative Reasoning: Enhancing Group Performance, Group Synchrony Modulation
Project code: SCI169
Supervisor(s):
Discipline(s):
School of Computer Science
Project
This project addresses the challenge of enhancing effective teamwork, which, despite its importance, is often hindered by cognitive overload, misaligned attention, and breakdowns in shared understanding within groups. The research aims to develop and investigate an AI agent designed to actively augment human collaborative reasoning by interpreting and responding to the real-time cognitive and physiological states of group members.
This AI agent will utilise physiological sensors alongside behavioural data to infer crucial cognitive states such as cognitive load, attention, engagement, confusion, and insight during collaborative tasks. Concurrently, it will assess various forms of inter-personal synchrony (behavioural, physiological, and inter-brain) known to correlate with effective collaboration.
Through reinforcement learning, the agent will learn to deliver subtle, adaptive interventions, such as tailored visual cues or auditory prompts, to optimise individual cognitive states, foster beneficial synchrony, and ultimately improve the group's problem-solving performance, efficiency, and the quality of their collaborative reasoning.
The role
The candidate will begin by comprehensively reviewing neuro-physiological markers of collaborative cognitive states, measures of inter-personal synchrony, reinforcement learning for human-AI interaction, and relevant experimental paradigms.
Subsequently, they will design and implement the experimental setup, including the collaborative task and neuro-physiological recording integration, develop machine learning models for real-time cognitive state and synchrony inference, design and train the AI agent, and conduct rigorous human-subject experiments to evaluate its effectiveness against baseline and simpler AI conditions.
The project culminates in analysing the rich dataset to understand the interplay between cognitive states, synchrony, AI interventions, and collaborative outcomes.
Automarker enhancement
Project code: SCI170
Supervisor(s):
Discipline(s):
School of Computer Science
Project
We want to improve the automated marking system that we use for algorithms courses.
Requirements
Some Linux, AWS, and web programming skills required.
Some tasks for improvement include multiple marking threads, more helpful feedback for incorrect submissions, database tools.
Visualizing the Sensorimotor Environments, Habitats and Coordinations of Two-Wheeled Virtual Agents
Project code: SCI171
Supervisor(s):
Discipline(s):
School of Computer Science
Project
This 10-week research project explores the concept of sensorimotor contingencies (SMCs), which describe how motor actions influence sensory experiences through interaction with the environment. To investigate this, the project uses simulations of Braitenberg vehicles—simple sensorimotor robots that react to environmental stimuli like light through hardwired sensor-motor connections. By observing how the structure and wiring of these vehicles shape the states they experience during movement, the project aims to formally define their SMCs using concepts from Buhrmann et al. (2013), distinguishing between the "sensorimotor environment" and "sensorimotor habitat."
The broader motivation for this work stems from theoretical perspectives by O’Regan and Noe, who argue that sensation arises not just from sensors, but from the structured interaction between sensory input and action.
The role
The project seeks to provide a minimal working example of how different robot designs lead to different repertoires of behavior and sensory experiences. This line of inquiry has implications not only for understanding perception and consciousness but also for identifying what current artificial intelligence systems lack—namely, embodied learning through sensorimotor experience, which humans begin developing from infancy.
Methodologically, the project will use Python to simulate the agents in a 2D light-based environment, employing tools from dynamical systems theory and machine learning. Agents will be variations of Braitenberg vehicles with dual sensors and motors, and their behavior will be analyzed using self-organizing maps to cluster and visualize SMC patterns.
Required skills
These include basic programming.
Using AI for Ophthalmology Triage
Project code: SCI172
Supervisor(s):
Discipline(s):
School of Computer Science
Project
AI has been used for various triaging applications in healthcare from identifying the severity of pancreatitis to identifying the best pathway when a person is admitted to hospital.
A significant portion of visits to ophthalmology emergency departments are for conditions that are not considered emergency. Through triage, people can be directed to the best care for their specific condition.
The role
This project will lay the ground for building a state of the art ophthalmology triage system for NZ. It will start with a scan of the landscape including a survey of the current research and tools, identify the specs for an ophthalmology system for NZ, and identify datasets that could be used to train a model for NZ.
Requirements
To undertake the project, you will require well developed research skills, good time management and an understanding of the application of AI.
The use of GenAI in Education
Project code: SCI173
Supervisor(s):
Discipline(s):
School of Computer Science
Project
Generative artificial intelligence (GenAI) has become increasingly prevalent in computer science education, influencing teaching and learning practices. With the advent of powerful GenAI models like ChatGPT and Gemini, educators and students are exploring new ways to integrate or mitigate the use of these tools in programming courses.
GenAI offers the potential as a valuable teaching aid, helping to explain complex concepts, generate exemplar code, and create personalised learning resources. However, its impact on education, particularly in tertiary-level computer science education, presents opportunities and challenges.
The role
We have administered a survey with tertiary Computer Science students to understand their usage of GenAI in education. This project will analyse the results of the survey, draw conclusions from the survey and produce a paper based on the survey.
Ideal student
We are looking for a student with good analysis and writing skills, together with an interest in the impact of AI on Education.
Intelligent Note-Taking and Key Event Detection for Enhanced Classroom Learning
Project code: SCI174
Supervisor(s):
Discipline(s):
School of Computer Science
Project
The integration of artificial intelligence into daily life has introduced numerous conveniences, and one particularly promising area of application is education, specifically, enhancing classroom efficiency. This project proposes the development of an AI-powered classroom assistant capable of automatically generating structured lecture notes and highlighting key classroom events, such as important explanations, student questions, and instructor emphasis.
By leveraging speech recognition, natural language processing (NLP), and event detection techniques, the system will analyze classroom audio (and optionally, video) in either real-time or through post-processing. The generated summaries aim to significantly improve students' review efficiency and overall learning experience.
The role
Candidates are expected to provide:
- A brief review of state-of-the-art NLP methods
- An overview of techniques for real-time data analysis
- A description of the application design
- An examination of existing annotated datasets relevant to classroom interactions and educational content
Building on this foundation, the student will apply advanced NLP technologies using publicly available educational datasets and lecture recordings. The goal is to develop a system that can generate concise, well-structured lecture notes from classroom audio in real time or after class.
As the project progresses, opportunities will arise to explore how specific classroom events, such as key explanations, student-instructor interactions, or topic transitions, can be automatically identified and reflected in the notes.
While the initial focus will be on audio-based input, the project may be extended to incorporate visual information (e.g., presentation slides or video recordings), depending on feasibility and available resources. The primary target domain will be undergraduate-level lectures (Arts preferences).
Required Skills
- Strong motivation and a willingness to learn
- Proficiency in Python programming
- Experience with NLP toolkits
Documenting Effective Practices for Software Co‑Development with Agentive AI
Project code: SCI175
Supervisor(s):
Discipline(s):
School of Computer Science
Project
Pair‑programming with LLM copilots is now commonplace, yet rigorous evidence on what actually works is sparse. This project prototypes an empirical study of best practices for joint human–AI software development and their distillation into a practical handbook.
The role
Key tasks include:
- Designing future controlled studies comparing coding productivity, defect rates, and developer satisfaction across interaction patterns
- Mining public GitHub repositories that document AI‑assisted commits to identify emergent conventions
- Synthesising findings into a living knowledge base and concise “Practice Patterns” guide for engineers and researchers
Skills gained
Participants will gain experience in mixed‑methods research, data mining, and technical writing; the resulting guide will be published through the Strong AI Lab and submitted to a software‑engineering conference and possibly serve as a basis for a new UoA course.
Extending KnowKat: A Consistent Interface for Accessing Knowledge Bases (“Ollama for KBs”)
Project code: SCI176
Supervisor(s):
Discipline(s):
School of Computer Science
Project
KnowKat is a prototype server that provides a uniform API that lets LLM agents query heterogeneous structured sources as if they were a single knowledge base.
This project will extend KnowKat into a plug‑and‑play “Ollama‑style” server for knowledge bases and their associated inference engines, adding:
- Hot‑swappable adapters for Wikidata, Neo4j, SPARQL endpoints, and domain‑specific SQL stores.
- Automatic schema alignment and entity‑resolution using embedding‑based similarity.
- Incremental caching and streaming responses to support low‑latency conversational agents.
Deliverables include a Dockerised micro‑service, integration examples with open‑source LLM stacks, and benchmarks on query accuracy and latency. Outstanding work may transition into a community‑maintained project and journal publication and opportunities for further research collaboration within the Strong AI Lab.
Automated Prompt and Instruction Design for Knowledge Acquisition by Specialised Sub‑Agents
Project code: SCI177
Supervisor(s):
Discipline(s):
School of Computer Science
Project
Large agent systems increasingly rely on sub‑agents—focused LLMs tuned for tasks such as citation extraction or hypothesis testing. Manually crafting high‑performing prompts for each sub‑agent is slow and brittle. This project develops an auto‑prompt‑engineering pipeline that learns optimal instructions for rapid knowledge acquisition.
The role
- Framing prompt design as a reinforcement‑learning or Bayesian optimisation problem with performance feedback signals.
- Building a library of reusable prompt templates parameterised by domain ontologies and task constraints.
- Demonstrating automatic generation of domain‑expert sub‑agents within the Strong AI Lab’s multi‑agent “Von” architecture and evaluating against hand‑written baselines.
Success will yield both publishable insights into LLM prompt optimisation and a practical toolkit that accelerates new agent creation across research projects. Excellent work may result in further research opportunities in the Strong AI Lab.
Visualising Internet service dependencies that are vulnerable to geohazards
Project code: SCI178
Supervisor(s):
Discipline(s):
School of Computer Science
Project
The VULGEO project is an international collaboration between Internet researchers from NZ, Europe, Japan, the US, Hong Kong and Australia as well as local volcanologist Shane Cronin from the School of Environment. It aims to identify which services users in NZ and the southwest Pacific might lose access to in case of a major geohazard event that takes out transoceanic submarine fibre cables.
The role
How many people would lose access to e-mail, cloud storage and other network-based resources? The team intends to find out – but how would we visualise what we find? This summer project will build a Google Earth based visualisation of the access paths to the services investigated.
Ideal student
The ideal student will have an interest in programming interactive applications, perhaps geographic information systems, and of course in networking!
Verification of network infrastructure geolocation
Project code: SCI179
Supervisor(s):
Discipline(s):
School of Computer Science
Project
The VULGEO project is an international collaboration between Internet researchers from NZ, Europe, Japan, the US, Hong Kong and Australia as well as local volcanologist Shane Cronin from the School of Environment. It aims to identify which services users in NZ and the southwest Pacific might lose access to in case of a major geohazard event that takes out transoceanic submarine fibre cables.
The role
How many people would lose access to e-mail, cloud storage and other network-based resources? Much of this work will use public databases of Internet infrastructure and services whose IP addresses have already been geolocated – but are these geolocators actually correct? Anycast, leased IP addresses and outdated information can mean that they are not where we think they are.
This summer research project will try and triangulate “known” services to confirm that we are identifying the right machines in the right locations.
Comprehensive Study on the Design and Applications of RISC-V's Vector Scalar Instructions
Project code: SCI180
Supervisor(s):
Discipline(s):
School of Computer Science
Project
This project aims to investigate the architectural design, operational principles, and practical applications of vector-scalar instructions within the RISC-V instruction set architecture. These instructions, part of the RISC-V Vector Extension (RVV), enable efficient parallel processing by combining scalar values with vector operands, offering significant performance benefits in domains such as machine learning, digital signal processing, and scientific computing.
The role
The project will involve a detailed study of the RVV specification, implementation of sample programs using vector-scalar operations, performance benchmarking against scalar-only approaches, and an evaluation of compiler and toolchain support. The ultimate goal is to assess the effectiveness and readiness of RISC-V vector-scalar instructions for real-world applications.
Skills Required
- Understanding of computer architecture and instruction set design
- Proficiency in RISC-V assembly and C/C++ programming
- Familiarity with Linux-based development environments
Prerequisites
- Coursework or experience in computer architecture and systems programming
- Basic understanding of vector processing and parallel computing concepts
- Comfort with low-level programming and debugging
Comprehensive Study on applying AMD Versal™" for AI Acceleration
Project code: SCI181
Supervisor(s):
Discipline(s):
School of Computer Science
Project
This project focuses on exploring the capabilities of AMD's Versal™ Adaptive SoC platform in accelerating artificial intelligence workloads. AMD Versal™ combines scalar processing, adaptable hardware (FPGA fabric), and intelligent engines (AI Engines) into a single heterogeneous computing platform, making it ideal for high-performance AI inference and edge computing applications.
The role
This project aims to investigate the architecture of the Versal™ platform, understand its AI Engine programming model, and evaluate its performance in accelerating AI tasks such as neural network inference. The study will involve setting up the development environment using tools like Vitis AI, implementing AI models on the Versal™ platform, and benchmarking their performance against traditional CPU/GPU-based implementations. The project will also explore real-world use cases in edge AI, computer vision, and embedded systems, assessing the trade-offs in power, latency, and throughput.
Skills Required
- Knowledge of digital design and embedded systems
- Familiarity with FPGA programming (VHDL/Verilog or HLS)
- Experience with AI/ML frameworks (e.g., TensorFlow, PyTorch)
- Proficiency in C/C++ and Python
- Understanding of heterogeneous computing and hardware acceleration
Prerequisites
- Coursework or experience in computer architecture and digital logic design
- Basic understanding of machine learning and neural networks
- Familiarity with Linux-based development environments
- Exposure to hardware-software co-design concepts
Full-Body Interaction with in a Virtual Reality Art Installation
Project code: SCI182
Supervisor(s):
Dr Becca Weber
Discipline(s):
School of Computer Science
Project
This project explores the future of interacting in VR – full-body interaction with AI agents within an immersive environment where users experience responsive audio-visual feedback based on real-time body tracking.
The role
We will work with Dance Movement Therapists to get design requirements for the existing software, which could be integrating AI agents into a Unity code, creating custom sound effects, creating custom visual effects. The research goal is to build a VR installation that supports embodiment and subjective experiences.
Over the summer, we will work with dance experts to iteratively develop the agents and interaction. This summer research project is related to a larger project that is likely to lead to topics of Masters and PhD studies and to collaborations with other researchers in universities abroad.
Skills required
Interest in VR development. Experience with PyTorch and/or Unity is a plus
Automatic assessment of accessibility, visual design, and interactivity of websites
Project code: SCI183
Supervisor(s):
Dr Becca Weber
Discipline(s):
School of Computer Science
Project
Web technologies are foundational and continue to be widespread, with front-end development skills in high demand. This project pursues the automatic assessment of web aspects dynamically, by executing them as would occur within a typical browser using the Selenium WebDriver framework.
The role
In this project you will write custom code to assess and interact with web components through browser-specific drivers, expanding functionality to assess visual Gestalt Principles and interactivity in programmatic fashion.
Selenium enables the remote control of a browser and mimics user actions on the browser including button click, drag, and drop selection, checkboxes, key presses, taps, and scrolling. The use of this tool is educational to support increased understanding of accessibility guidelines and visual design skills.
We will build out documentation and resources in our aim to provide this tool to other educators. There are options to explore how machine learning and/or classifiers and/or LLM can support student learning of visual design and accessibility in web technologies.
Fall Detection-Enabled Risk Management for Aged Care
Project code: SCI184
Supervisor(s):
Discipline(s):
School of Computer Science
Project
Dementia is a complex and progressive neurodegenerative condition. Individuals living with dementia often exhibit wandering behaviour in rest homes or care facilities. While this behaviour may provide a sense of autonomy, it also poses significant safety risks, particularly the risk of falls.
The role
To address these challenges, this project aims to develop an advanced indoor activity tracking system that enhances safety, supports caregivers, and preserves the dignity of individuals with dementia.
Building on an existing Ultra-Wideband (UWB)-based system, this project will expand its capabilities by integrating fall detection and advanced risk management features tailored for indoor environments in care facilities. The system will provide continuous, high-precision monitoring of movement patterns, enabling early risk detection and rapid response to incidents. Through comprehensive evaluation, the solution will be refined to maximize its effectiveness in reducing the risks associated with indoor wandering.
This project will enhance the current UWB-based system by incorporating fall detection and other advanced risk management functionalities for indoor activity tracking in care facilities. The intellectual property (IP) associated with these software extensions will be owned by the funded research project.
Leveraging LLMs for Automated Legacy Software Refactoring and Upgrade
Project code: SCI185
Supervisor(s):
Discipline(s):
School of Computer Science
Project
This project explores the use of Large Language Models (LLMs) to automate the modernization of legacy software systems. Legacy codebases are often poorly documented, difficult to maintain, and incompatible with modern technologies.
The role
By leveraging the code understanding, transformation, and generation capabilities of LLMs, this project aims to develop an intelligent upgrade framework that can analyze outdated code, recommend improvements, refactor components, and assist in migrating systems to newer platforms or architectures. The approach integrates static analysis, prompt engineering, and validation techniques to ensure reliability, maintainability, and compliance with contemporary software standards. The goal is to reduce the cost and risk of system modernization while accelerating digital transformation across industries.
Auditing Artificial Intelligence with Adversarial Learning
Project code: SCI186
Supervisor(s):
Discipline(s):
School of Computer Science
Project
We aim to design and develop new methods to attack machine learning models and use the adversarial attacks to define a measure of reliability. Weak performances of models where data sets are not representative or flaws in training process are a common issue in Machine Learning. This leads to misclassification and unfairness of the model.
The role
We will develop a framework that identifies adversarial regions in the data space that are prone to make models fail. The framework will not only identify these regions and data, but also produce tools to improve it, and return a score that reflects the reliability of the model. This score can be used to certify models without having access to the training process and estimate the applicability of models to specific use cases.
Recommended skills
Basic knowledge of machine learning and Python or Rust
AI Prediction of Persistence of Environmental Pollutants
Project code: SCI187
Supervisor(s):
Discipline(s):
School of Computer Science
Project
Most chemicals that are currently produced sooner or later end up in the environment, many of them in rivers and other waters. It is essential to know their fate in terms of transformations and persistence. Harmful chemicals that degrade quickly might pose no big thread to the environment, however persistent toxic compounds can have lasting negative impact.
The role
In this project, we will improve methods to predict the environmental fate of pollutants and chemical transformations. We will develop novel AI methods that predict complex pathways and learn the chemical processes behind it.
Recommended skills
Basic knowledge of chemistry, machine learning, and Python
Design for Degradability - AI-based Development of Sustainable Chemicals
Project code: SCI188
Supervisor(s):
Discipline(s):
School of Computer Science
Project
An important aspect in the development of novel chemicals is their environmental fate, that is their ability to degrade when released in the environment. To achieve this, the goal is to design compounds that fulfil a certain function, for example medication or pesticides, and at the same time allow for quick degradation into harmless metabolites.
The role
We will develop new algorithms that achieve this, evaluating on large databases of existing compounds. We will use AI models for predicting degradation products and pathways (see enviPath - https://envipath.org). Our approach will be to start with existing compounds, and transform them using adversarial methods and generative models (GANs) such that their biodegradability increases while at the same time keeping their original function.
Recommended skills
Basic knowledge of chemistry, machine learning, and python
Adversarial Time Series
Project code: SCI189
Supervisor(s):
Discipline(s):
School of Computer Science
Project
Adversarial Machine Learning is a field of Machine Learning that focuses on exploiting model vulnerabilities by making use of obtainable information from the model. Studying a model’s weaknesses to adversarial attacks not only helps the researcher understand more about the model itself, but also allows them to defend against malicious attacks and prevent potentially fatal consequences after deployment.
Adversarial Machine Learning was firstly proposed in the image classification domain, where an attack fools a model to misclassify an image by adding carefully crafted noise that is hardly detectable by a human. Recently, adversarial methods have been introduced that target time series challenges. We will develop and evaluate new adversarial attacks on time series, targeting specific time series challenges beyond forecasting.
Recommended skills
Basic knowledge of machine learning and Python
I-based Retrosynthesis
Project code: SCI190
Supervisor(s):
Discipline(s):
School of Computer Science
Project
Retrosynthesis is a fundamental concept in chemistry and materials science that involves working backwards from a target molecule to identify the sequence of reactions needed to synthesize the target compound. Revealing synthesis pathways that should have been documented (but were not) is a handy tool. However, substantially more impactful is identifying alternative sequences that use cheaper, more environmentally friendly, or otherwise superior materials, involve fewer steps, enhance yields, mitigate risks, etc.
The role
This project will review and enhance or develop AI-based retrosynthesis methods.
Recommended skills
Basic understanding of chemistry, machine learning, and Python
Using AI To develop a method for injury prevention in professional sports
Project code: SCI191
Supervisor(s):
Discipline(s):
School of Computer Science
Project
Modern live sports broadcasts incorporate many analytics and graphics to improve the audience's viewing experience and provide on-time information to coaches. Some applications still evade teams such as the prevention of injuries based on video footage of a given player. This may be based on unusual movements or behaviour of the player(s) as captured in publicly available video footage.
The role
This project will first review state-of-the-art methods and apply methods to various publicly available datasets. The student will then create their own dataset of injuries based on publicly available NZ sports footage and use the wealth of publications and expertise within the IVSLab (post-doc, PhDs, Masters) to produce and test a novel method based on traditional machine learning methods or a combination of traditional and deep learning-based approaches.
Data
This project will use free publicly available datasets from various sports, including but not limited to the following three groups:
i) basketball/netball
ii) rugby league/soccer
iii) rugby union).
Desired Output
- A literature review of state-of-the-art methods for injury detection/description in sports games.
- A comparison of two or three existing (published) methods, and potentially a new method for “injuries detection before they happen” using sports video footage.
- Well-documented code following standards set by the supervising team.
- An extensive Report (in LaTeX/Overleaf).
Who are we looking for?
Someone with experience with deep learning and an interest in learning required computer vision basic concepts.
Available Resources
Access to GPUs or PCs will be provided as needed.
Constraints
- Weekly reporting with the supervisors.
- Weekly attendance and presentation to the IVSLab meeting.
- The student will need to go through extensive video footage and produce the relevant annotations using CVAT software (available in the IVSLab).
- Must use Overleaf (LaTeX) for reporting and sharing progress.
Automated Detection of Intertidal Macrofauna and Sediment Surface Features from Low-Tide Imagery
Project code: SCI192
Supervisor(s):
Patrice Delmas (ComSci)
Shahrokh Heidari (IMS)
Discipline(s):
School of Computer Science
Project
Intertidal macrofauna play a key role in shaping sediment dynamics and ecosystem functioning. Organisms such as worms, anemones, and shrimp leave distinctive surface features visible during low tide. These features can be used as ecological indicators of species presence, activity, and spatial distribution, providing important insights for benthic habitat assessment and long-term environmental monitoring.
The role
This project aims to develop a computer vision-based system to automatically detect and count a selected set of intertidal macrofauna features from low-tide imagery. The project will involve a field trip to Leigh Marine Laboratory to collect imagery using standardized survey protocols.
The student will:
- Participate in fieldwork at Leigh to acquire low-tide imagery
- Perform semi-automated annotation using provided tools and guidance
- Train and evaluate a detection model on the annotated dataset
- Generate spatial summaries and present findings
Skills gained
This interdisciplinary project offers hands-on experience in data collection, annotation, AI model development, and ecological interpretation, contributing to emerging methods for scalable and non-invasive benthic monitoring.
Skills required
Strong motivation and interest in marine science or environmental monitoring. Basic familiarity with Python and PyTorch is desirable. Prior experience in computer vision or machine learning is not essential, but the candidate should be willing to learn relevant tools during the project.
Developing an Unsupervised or Semi-supervised pipeline to Segment Underwater Algae in New Zealand
Project code: SCI193
Supervisor(s):
Discipline(s):
School of Computer Science
Project
Underwater algae monitoring is crucial for evaluating the health of marine ecosystems and mitigating environmental risks. However, different regions or countries have different types of algae, making a general algae detector and monitoring system development very challenging. Some algae in New Zealand may have different shapes or colours, thus, a deep learning model trained on other types of algae may not work with the species in New Zealand.
Annotating algae can also be time-consuming and challenging, as some species may have numerous leaves and branches. Without a proper ground-truth label of the algae, supervised learning methods can be difficult.
The role
In this project, students should aim to utilize any unsupervised or semi-supervised methods to develop a pipeline that can be used to segment underwater algae specifically found in New Zealand.
Skills required
Strong motivation and a willingness to learn. Some Python programming capabilities and the ability to pick-up existing tools and knowledge. Computer vision knowledge is not essential although the candidate will need to pick-up relevant skills in computer vision and machine learning along the way.
Concrete Computational Complexity
Project code: SCI194
Supervisor(s):
Discipline(s):
School of Computer Science
Project
One of modern mathematics' greatest problems is whether P equals NP or. We seem far away from answering this question, but we are able to prove that some problems are hard for limited computational models. For example, if we give a number to two separate parties, it is impossible for them to know if they hold the same number without revealing the whole number. The project consists of exploring similar hardness results.
The role
The first part of the project consists of the candidate acquainting themselves with one weak computational model such as pebble games, decision trees, communication protocols, propositional proofs, or monotone Boolean circuits. In other words, survey the cornerstone results and research techniques used to prove hardness results in that computational model. This will involve reading books and research articles with guidance from the supervisor.
The second part consists of applying these techniques to attempt to prove that some particular task cannot be solved efficiently in the model of choice. This will involve research discussions with the supervisor consisting of brainstorming ideas and discussing technical roadblocks, developing ideas and formalising them into proofs individually, and writing down finished proofs. An example of a problem that may be attempted during the second part is designing a graph whose black pebbling number and reversible pebbling number differ.
Skills required
Mathematical maturity, including the ability to read and write formal proofs, and a strong background in discrete mathematics. Prior knowledge of theory of computation is strongly recommended but not essential. Programming experience is not required.
AI for Marine Monitoirng: Detection of sea urchin barren reefs from underwater imagery
Project code: SCI195
Supervisor(s):
Co-supervisor: Arie Spyksma (Institute of Marine Science)
Discipline(s):
School of Computer Science
Project
Kelp forests are among the most productive ecosystems on Earth, but climate-driven impacts are causing wide-spread kelp habitat loss. For example, the climate-driven proliferation of the longspined sea urchin is one of the most urgent threats to kelp forests in south-eastern Australia and north-eastern New Zealand.
Assessing this threat requires the collection and analysis (typically manually) of underwater imagery spanning tens to hundreds of kilometres of reef. The high contrast of sea urchins on barren reef makes this an ideal candidate for modern computer vision solutions based on machine learning (ML) algorithms to dramatically improve annotation and analysis.
The role
Using existing image-based monitoring data you will co-develop and test ML algorithms to detect the presence and the extent of urchin barren expansion in Australia/New Zealand.
Ideal student
This project is suitable for students with basic skills in maths, statistics, machine learning and image analysis; intermediate programming skills in Python. Familiarity with convolutional neural networks and programming experience in Pytorch will be beneficial (but it is not necessary and can be learned while working on the project).
Reliable machine learning for regression problems
Project code: SCI196
Supervisor(s):
Discipline(s):
School of Computer Science
Project
Large datasets are now routinely collected from diverse sensing technologies that need to be integrated and analysed in an efficient and intelligent way in order to support decision-making in almost in any domain of our lives, spanning from medicine, to environment, and transportation.
While deep learning models have seen enormous success in computer vision due to their high expressiveness compared to traditional shallow models, they don’t have well-motivated methods for accurately estimating the uncertainty in their prediction.
The role
In this project you will get familiar with and investigate methods for quantifying uncertainty in deep learning model predictions for regression problems, when your target of interest is a continuous variable as in time series analysis (e.g. predict wind speeds or weather temperature at a given location)
Requirements
This project requires basic knowledge of statistics, machine learning, and good knowledge of Python. While it is beneficial to have essential understanding of deep learning networks, it is not necessary and can be learned while working on the project.
Reliable machine learning for object identification
Project code: SCI197
Supervisor(s):
Discipline(s):
School of Computer Science
Project
The research project is focused on improving the confidence calibration of object detection models in images. While modern object detectors have achieved impressive performance, their confidence scores are often miscalibrated, leading to overconfident or underconfident predictions. This in simple words means that you cannot trust the model, which can be especially problematic in safety-critical applications such as autonomous driving or cancer detection.
The role
The goal of this project is to analyse, benchmark, and improve the reliability of confidence estimates in popular object detectors. You will explore existing calibration methods, evaluate performance on standard benchmarks, and develop new approaches.
Ideal student
This research is ideal for students interested in computer vision, deep learning, and reliable machine learning.
An online authoring tool for individualized paper-based examinations
Project code: SCI198
Supervisor(s):
Discipline(s):
School of Computer Science
Project
Paper-based examinations are more secure than their digital counterparts. We use a framework called Dividni, which offers a command-line interface (CLI) for authoring individualized paper-based exams—meaning that every student receives a unique version of the exam, with different questions (and answers) from their peers.
The role
Recognizing that CLIs are not user-friendly, this project aims to develop a professional online frontend for creating exam papers, using the CLI in the backend (see tutorial: https://dividni.com/tutorial).
Required skills
To succeed in this project, you should have excellent web development skills and a passion for teaching and learning. While some backend development in C# will be required to support the CLI, the majority of the effort is expected to focus on the frontend.
Pacman Assembly Arena: A Web-Based Assembly Programming and Debugging Tool
Project code: SCI199
Supervisor(s):
Discipline(s):
School of Computer Science
Project
This project aims to develop Pacman Assembly Arena, an engaging, web-based educational platform designed to teach assembly programming through interactive gameplay.
The role
Students will write assembly code to control characters such as Pacman or ghosts, gaining hands-on experience with low-level programming concepts in a fun and visually rich environment.
The student will contribute to the design and development of key platform features, including a visual debugger, step-by-step code execution, and a game engine that interprets and runs student-authored assembly instructions in real time.
Skills developed
This project offers a unique blend of front-end and back-end web development, interactive game design, and low-level programming. It is particularly well-suited to candidates interested in both computer architecture and educational technology.
Skills required
Motivation and creativity, with a willingness to learn. Some experience in web development (e.g., JavaScript, HTML/CSS) and an interest in systems or low-level programming. Familiarity with game logic and debugging tools is helpful but not essential.