Global Taste of Research - India Summer Practicum Scholarship Program UGVC1062 Research Projects

Project Title: Investigating the immune response to medical implants using novel genomics tools

Supervisor(s): Nona Farbehi

Email: n.farbehi@unsw.edu.au

Description: The integration of biomaterials into host tissues is influenced by their physical properties, including shape and porosity, which dictate cell recruitment, immune response, and tissue remodeling. Despite extensive research on biomaterial design, the spatial organization of cells around implants with varying morphologies remains poorly understood. This study introduces and investigates the impact of implant shape and porosity on cellular interactions using spatial transcriptomics to map gene expression and cell migration patterns.

To assess cellular response, implants were implanted subcutaneously in mice, and spatial transcriptomic analysis was performed at multiple timepoints post-implantation. Gene expression patterns demonstrated distinct cellular organization around different microgel shapes. The Visium dataset further highlighted differential immune activation and tissue integration, providing new insights into the biomaterial–host interface. We need students with a bioinformatics background to analyse the spatial data. These insights will inform the rational design of implants for regenerative medicine and tissue engineering applications.

Project Title: High throughput screening of biomaterial interactions with the vascular endothelium: towards next-generation drug delivery systems

Supervisor(s): Jonathan Yeow; Megan Lord

Email: j.yeow@unsw.edu.au; m.lord@unsw.edu.au

Description: In recent years, nanoparticle carriers for drug delivery have been proposed as a promising approach to enhance the therapeutic index of medications. Although researchers have made significant progress in designing and engineering nanoparticle carriers, it is estimated that only a very small fraction (< 2 %) of the injected nanomedicine actually reaches the intended site resulting in limited translation of so-called nanomedicines into the clinic. One reason for this low efficiency is the endothelial glycocalyx, a dynamic gel-like coating rich in sugars (glycans) that lines the blood vessel wall and is increasingly being recognised for its role as a cellular gatekeeper that regulates interactions of nanomaterials delivered via the blood stream.

In this project, you will utilise high throughput synthetic and characterisation tools to screen how libraries of nanomaterials interact with in vitro models of the endothelial glycocalyx. This will enable the development of key structure-activity relationships, greatly enhancing our understanding of how to optimise biomaterials for more efficient delivery via the endothelium.

Project Title: AI for Multimodal, Multispectral Imaging of Blood Cells in Neurodegenerative Diseases

Supervisor(s): Akanksha Bhargava

Email: a.bhargava@unsw.edu.au

Description: The project focuses on applying artificial intelligence (AI) methods to multidimensional imaging datasets of blood cells collected from healthy and disease cohorts. The student will work with high-content imaging data and develop computational approaches to extract quantitative features, identify disease-relevant patterns, and build classification or predictive models for monitoring disease and response to treatment. The work is conducted in partnership with Macquarie Hospital Sydney and contributes to the development of an early stage startup venture for Motor Neurone Disease (MND). Key components may include: Image preprocessing and segmentation (multispectral imaging data), Feature extraction from multimodal imaging data, Machine learning and/or deep learning model development, Model evaluation and biological interpretation.

The project provides exposure to translational biomedical research at the interface of engineering, data science, and clinical medicine. The student will gain experience working with real-world clinical datasets and contribute to the development of AI-driven diagnostic technologies. Desired Skills: Programming experience in Python or MATLAB; Familiarity with image analysis; Knowledge of machine learning or deep learning (preferred); Interest in biomedical engineering, AI in healthcare, or translational research. "

Project Title: Computational design of Na battery cathodes for sustainable energy storage

Supervisor(s): Priyank V Kumar

Email: priyank.kumar@unsw.edu.au

Description: This project aims to use computational methods to design and improve cathode materials for sodium (Na) batteries for sustainable energy storage. Na batteries are attractive because sodium is abundant and low cost, but their cathode materials often suffer from low energy density and poor long-term stability. In this project, atomistic electronic structure calculations and machine learning methods will be used to study Na cathode materials. The project will also explore doping and elemental substitution strategies to improve capacity, stability, and rate performance.

The results will provide clear design guidelines for selecting and optimizing Na battery cathodes, reducing experimental trial and error and supporting the development of low-cost and durable energy storage systems.

Project Title: Sequence informed solubility modelling of edible plant-based proteins

Supervisor(s): Rishi Ravindra Naik

Email: r.naik@unsw.edu.au

Description: Protein solubility is one of the most critical functional properties in food systems, directly influencing texture, emulsification, foaming, digestibility, and product stability. In plant-based and alternative protein development, limited solubility, particularly near the isoelectric point often restricts ingredient performance and application versatility. Despite its importance, solubility is typically determined experimentally through trial-and-error formulation rather than predicted from molecular properties. This project aims to build and experimentally validate a simple, sequence-based solubility prediction model for proteins. Using only primary amino acid sequence information (FASTA), the project will focus on computing basic physicochemical descriptors such as hydropathy (GRAVY) and net charge as a function of pH . These descriptors will be combined into a minimal predictive model that estimates relative protein solubility across a pH range.

The experimental component will involve measuring apparent protein solubility (%) at different pH values using a standardized centrifugation and Bradford assay protocol. Predicted solubility trends from the model will then be compared with measured data to assess whether simple sequence-derived properties can capture first-order solubility behaviour (e.g., reduced solubility near the isoelectric region and increased solubility away from it).

Project Title: Engineering a next-generation zinc battery with improved durability, manufacturability, and recycling for stationary energy systems

Supervisor(s): Dipan Kundu

Email: d.kundu@unsw.edu.au

Description: This project aims to deliver a safe, affordable, and long-lasting battery prototype to support Australia’s clean energy transition. Building on breakthroughs made over the last few years, the team will develop a zinc-ion battery MVP that is recyclable, non-flammable, and cost-effective - ideal for residential and grid-scale energy storage. The project will improve battery performance, test durability under harsh conditions, and prepare for manufacturing and commercialisation. 

Project Title: Fate of emerging contaminants in the environment

Supervisor(s): Denis O'Carroll

Email: d.ocarroll@unsw.edu.au

Description: Contaminants of emerging concern (e.g., PFAS, bisphenol-A, 6PPDQ) have negative impacts to human and ecosystem health (e.g., carcinogens, toxicity to fish) as well as potential to develop antimicrobial resistance. There is a limited understanding of how they travel and react in our surface and groundwaters. This project will improve our understanding of the mechanisms that control contaminants of emerging concern in our waters. The inadequate understanding of where contaminants of emerging concern go and react in the environmental has stymied planning and implementation of a number of high-profile public infrastructure projects.

These delays and public uncertainty can result in a loss of $100s of millions in taxpayer money. This project will help to limit the negative impacts of contaminants of emerging concern by improving our understanding of contaminants of emerging concern fate. The project team has the expertise to conduct their crucial work and an extensive network to work to disseminate outcomes to the public, academic community and regulators.

Project Title: Behavioural Analytics & Scenario Insights System

Supervisor(s): Taha Rashidi

Email: z3423089@unsw.edu.au

Description: This project aims to develop a public transport model for Western Sydney. The research project contributes to our broader multimodal model for the Western Sydney Area. The demand model links to a traffic assignment model to assess the impact of pricing, accessibility, and level of service on mode shift. 

Project Title: Developing sustainable graded porous structures

Supervisor(s): Daniel Chen

Email: daniel.chen7@unsw.edu.au

Description: Innovations in composite materials drive advances in engineering. Introducing internal porosity into traditionally homogeneous matrices has led to a new class of multifunctional, lightweight composite structures. These porous forms not only reduce structural weight but also offer enhanced performances, including improved energy absorption, superior thermal regulation, and distinctive vibration characteristics. Such advantages make porous structures highly attractive for a wide range of industrial applications. This project aims to pioneer innovations in green civil engineering by developing novel porous cementitious structures tailored for sustainable built environment.

The research will establish graded porous geometries that simultaneously enhance structural stiffness and thermal insulation while minimizing material usage. These lightweight yet robust designs are critical for reducing cement consumption, thereby mitigating carbon emissions associated with both construction and building operation, and contributing to the national Net Zero Plan.

Project Title: Quantum computing for resilient urban transport networks

Supervisor(s): Divya Nair

Email: divya.nair@unsw.edu.au

Description: This project explores how quantum computing techniques can improve the resilience and efficiency of urban transport systems. It focuses on identifying critical network structures, congestion bottlenecks, and risk-prone link combinations using advanced optimisation methods. Students will work with real traffic data, build and analyse optimisation models, and gain practical experience at the intersection of quantum computing and transport engineering.

Project Title: From digital geometric models to fully-automated adaptive analysis of structures and materials

Supervisor(s): Chongmin Song

Email: c.song@unsw.edu.au

Description: Modern digital geometric models, such as digital imaging and point clouds, are increasingly being applied in the design and manufacturing in the fields of civil, mechanical, material, biomedical and other disciplines of engineering and science. Conventional computational methods require great human efforts to utilize these geometric models for engineering analysis. This project aims to develop a full automatic approach to perform advanced engineering analysis of materials and structures directly from various formats of digital models.

Work could involve the development of novel algorithm for the creation of numerical models from point clouds, digital images and stereolithography models, adaptive stress analysis using the scaled boundary finite element method, modelling of soft materials, and development of graphical user interfaces for pre- and post-processing. Computer code for automated analysis will be developed on a high-performance computing facility. Strong analysis and computing skills are required for this project. Experience with high-performance and cloud computing is high desirable. 

Project Title: Design optimisation and experimental validation of flexo-electric meta-materials for self-powered micro- and nano-sensors

Supervisor(s): Elena Atroshchenko, Aron Woldeghiorgis (Michael)

Email: e.atroshchenko@unsw.edu.au

Description: The next generation of intelligent infrastructure, personalized healthcare devices, and wearable technologies will rely on dense networks of micro- and nano-scale sensors that must operate without external power sources. This requirement makes efficient energy harvesting from the surrounding operating environment essential for achieving long-term, maintenance-free sensor operation. However, existing energy harvesting technologies lack predictive design frameworks that can capture size-dependent electro-mechanical coupling and enable systematic optimisation of micro- and nano-scale devices.

To address this challenge, this project will establish an advanced numerical modelling and optimisation framework for designing and implementing novel, highly efficient flexo-electric meta-material-based Energy Harvesting and Sensing (EHS) system at the micro- and nano-scale. The novelty lies in combining (i) advanced coupled electro-mechanical models based on strain-gradient elasticity with flexo-electricity, capturing size effects and strain-gradient-induced polarization at the micro/nano-scale; (ii) advanced shape and topology optimisation methods, based on isogeometric analysis, for the design of meta-material architectures of two types: locally resonant and auxetic lattices; (iii) voltage-based sensing framework for efficient monitoring of parameters of interest; (iv) Uncertainty Quantification (UQ): Bayesian inference to extract hard-to-measure flexoelectric and gradient parameters from experiments, ensuring predictive accuracy and robust designs; and (v) prototyping and experimental demonstration of an EHS system at micro/nano-scale.

The project uniquely integrates advanced numerical modelling and optimisation expertise at the School of Civil Engineering with state-of-the-art microfabrication and experimental characterisation capabilities at the MEMS laboratory in the School of Electrical Engineering. By closing the loop between modelling, optimisation, uncertainty quantification, and experiment, the project will enable physics-driven engineering of next-generation micro-scale device technologies.

Project Title: AI for resilient urban transport networks

Supervisor(s): Divya Nair

Email: divya.nair@unsw.edu.au

Description: This project examines how artificial intelligence can enhance resilience analysis and decision-making in urban transport networks by identifying critical structures, congestion patterns, and time-dependent connectivity changes using data-driven methods. It combines traditional transport models with large-scale traffic data and AI techniques, while engaging students in data processing, modelling, analysis, and visualisation to build research and analytical skills.

Project Title: Precision-configurable Photonic Computing for Accelerating AI Workload

Supervisor(s): Haibo Zhang

Email: haibo.zhang@unsw.edu.au

Description: The rapid growth of artificial intelligence (AI) has placed increasing demands on computational throughput and energy efficiency. Conventional electronic processors face fundamental challenges in meeting these demands due to power density limits, memory bandwidth bottlenecks, and scaling constraints. Photonic computing has emerged as a promising alternative, offering high bandwidth, massive parallelism, and low-latency signal propagation. However, most existing photonic computing systems are optimised for low-precision or approximate computation, which limits their applicability to workloads requiring high numerical accuracy.

Achieving high-precision computation in photonic systems remains challenging due to analogy noise, device nonlinearity, quantisation effects, and limited analogy resolution. This project aims to develop algorithmic techniques for enabling high-precision photonic computing, with a focus on precision-configurable designs that balance numerical accuracy, energy efficiency, and hardware complexity. This project demands in-depth knowledge of reconfigurable computing, as well as extensive hands-on experience with FPGA design and Python programming. 

Project Title: Privacy Risks in Conditional Generative Models

Supervisor(s): Salil Kanhere

Email: salil.kanhere@unsw.edu.au

Description: Generative models extend far beyond large language models like ChatGPT. In computer vision, diffusion models are the current state of the art for image generation, while GANs remain highly effective. A key ingredient behind their success is conditional generation, guiding the model with additional information to improve quality, stability, and control. In images, conditioning may specify whether a person is wearing glasses or whether the style should be realistic or artistic. In trajectory and structured data generation, conditioning on context or attributes is often essential to produce meaningful outputs. Generative models are increasingly used to create privacy-preserving synthetic data as a substitute for sensitive datasets. While training data extraction is difficult, privacy leakage remains a known risk.

However, an important and underexplored question is: Does conditioning during generation introduce new privacy threats? By reducing uncertainty and narrowing the output space, conditional inputs may increase the risk of leaking information about individuals in the training data, especially for rare or sensitive attributes. This internship project will investigate whether and how conditional generation amplifies privacy risks in modern generative models. Through a well-defined threat model and empirical evaluation (e.g., membership or attribute inference attacks), the goal is to quantify the privacy impact of conditioning and provide insights for safer deployment of generative models in sensitive domains.

Project Title: Privacy Risks from Long-Term Memory in Autonomous LLM Agents

Supervisor(s): Salil Kanhere

Email: salil.kanhere@unsw.edu.au

Description: Autonomous LLM agents increasingly rely on persistent memory to improve personalisation, continuity, and task performance. These systems store conversation histories, retrieved documents, summaries, and user-specific context across sessions. While such memory mechanisms enhance usability, they may also introduce new and underexplored privacy risks. This project investigates whether long-term memory in autonomous LLM agents can lead to unintended information leakage across users or sessions. Specifically, we study whether sensitive data provided by one user can be extracted (directly or indirectly) by another user through carefully crafted prompts or reasoning chains.

The internship project will involve implementing an agent framework with different memory configurations (e.g., short-term context, summarised memory, vector-store retrieval) and designing systematic extraction attacks to evaluate leakage risks. We will quantify privacy exposure under varying memory architectures and threat models, and analyse the trade-off between agent performance and privacy guarantees. The outcome will be a structured threat model, an empirical evaluation of memory-induced privacy risks, and recommendations for safer deployment of autonomous LLM agents in real-world applications.

Project Title: Methods for formally reasoning about programs

Supervisor(s): Vineet Rajani

Email: v.rajani@unsw.edu.au

Description: The project will broadly be in the area of logic and verification alongside considering their applications to domains like trustworthy AI and security/privacy. Depending on the interest of the student, possible research directions include building new methods (e.g. type theories or programs logics) to reason about interesting properties (e.g. pertaining to security or privacy, secure and verified compilation, complexity/cost analysis) of programs/models (e.g. probabilistic, neural, quantum), including their mechanisation in a (interactive) theorem.

Project Title: Diffusion Reason

Supervisor(s): Flora Salim

Email: flora.salim@unsw.edu.au

Description:  This project explores how to leverage the generative power of diffusion models to improve systematic reasoning in challenging abstraction tasks (e.g., IQ-test-like puzzle tasks). We combine diffusion-based generation with modern reasoning techniques (e.g., hierarchical decomposition, iterative refinement, verification/self-consistency, and reasoning modules) to produce solutions that are more robust, compositional, and generalisable from limited examples. The project will lead to development of robust and efficient reasoning models and a research paper on this topic.

Project Title: Verifiable Credentials for Digital Identity Management

Supervisor(s): Sushmita Ruj

Email: sushmita.ruj@unsw.edu.au

Description: Credentials are integral to digital identity and to our daily lives: driver's licenses confirm our capability to operate motor vehicles; university degrees assert our level of education; and government-issued passports attest to our citizenship when traveling between countries. Establishing an adequate Digital ID infrastructure can reduce the need to repetitively hand over sensitive identity documents to multiple organisations. Verifiable credentials are digital credentials that provide a set of tamper-proof claims and metadata that can be verified for authenticity. Verifiable credentials (VC) are digital credentials that follow the relevant World Wide Consortium Open standards.

By using Cryptography and privacy enhancing technologies (PETS), VCs have the power to verify a user’s assertion without revealing sensitive information. Anonymous credentials are used to verify ownership of credentials without disclosing their privacy. The project aims to explore recent advances in anonymous verifiable credentials and their applications. The aim is to build anonymous verifiable credentials that are unlinkable (two presentations will not be linked to the same user), discloses minimal information, and can be revoked when required.

The project involves both theory and implementation. The project involves excellent understanding of data structures, discrete mathematics, and algorithms. This space is highly active and can lead to impactful results. References: - https://www.w3.org/TR/vc-data-model-2.0/ https://www.nist.gov/video/stppa4-talk-1-anonymous-credentials https://link.springer.com/chapter/10.1007/978-3-031-30731-7_6 

Project Title: Agentic Sensor Language Models for Reasoning over Time-Series

Supervisor(s): Flora Salim

Email: flora.salim@unsw.edu.au

Description: Recent efforts applying LLMs for time-series analysis and classification tasks, typically by converting time-series signals to text or images to be processed by LLMs or VLMs, suffer from poor performance and lack of interpretability. We have pushed the boundaries of using LLMs for time-series classification through SensorLLM and ZARA. In Sensor LLM, we introduced a two-stage framework—first aligning sensor data with intuitive trend descriptions, then fine-tuning for classification—achieving accuracy on par with or surpassing state-of-the-art supervised models. In contrast, ZARA eliminates retraining altogether, proposing an agent-based, zero-shot framework that combines a feature knowledge base, evidence retrieval, and hierarchical reasoning to predict activities while also providing natural-language explanations.

While SensorLLM highlights generalization through human-intuitive alignment and supervised finetuning, ZARA emphasizes interpretability and plug-and-play adaptability, achieving strong zero-shot performance and explainability. This Taste of Research project will build on our past research. The focus of the project is on verifiable reasoning, since most current reasoning/thinking lacks explicit evidence support. The overarching goal is to enhance time-series reasoning capabilities over a broad time-series tasks, enabling robust QA and chat-based exploration over high dimensional time-series. 

Project Title: Spatiotemporal Reasoning Challenge

Supervisor(s): Flora Salim

Email: flora.salim@unsw.edu.au

Description: The project focuses on the Spatiotemporal reasoning challenge, which focuses on benchmarking LLMs and reasoning model capabilities on spatiotemporal prediction and reasoning tasks, for applications in cyber-physical systems (CPS) such as robotics, autonomous vehicles, and smart city infrastructure. Tasks include benchmarking LLM and LRM performance in geometric localization, trajectory tracking, spatial/temporal relationship inference, and real-world navigation tasks. Experiment settings and evaluation can vary, such as through question-answering and/or through  multi-step reasoning pipelines in simulated agentic environments.

This project is appropriate for coursework students as it provides a well-defined and structured challenge with clean datasets and clear research objectives. It encourages students to come up with their own problem-solving strategies to tackle spatiotemporal reasoning tasks. Additionally, the project serves as an excellent opportunity for students to develop independent research and communication skills, providing a strong foundation for those interested in pursuing research studies in the future. Techniques to be explored include: time-series forecasting, spatiotemporal reasoning, LLM-based approaches, agentic AI.

Project Title: RF‑Based Activity Monitoring Using Wi‑Fi CSI for Contactless Health and Behaviour Assessment

Supervisor(s): Deepak Mishra

Email: d.mishra@unsw.edu.au

Description: This internship project aims to develop a lightweight RF‑sensing framework that uses Wi‑Fi Channel State Information (CSI) to monitor and classify human activities in an indoor environment. Leveraging ubiquitous Wi‑Fi signals, the system extracts CSI amplitude variations to infer motion dynamics without requiring cameras or wearable sensors, enabling a scalable, privacy‑preserving sensing solution.

The three‑month internship will focus on CSI data preprocessing, noise filtering, and feature extraction, followed by the implementation of efficient machine‑learning models to classify basic activities such as walking, sitting, standing, and transitions. Agya Sanghi, an outstanding IIT Delhi undergraduate with strong foundations in electronics, programming, and applied engineering, who approached me late last year, will be selected as the intern for this project. Herstrong background in electronics, data structures, C++ and Python programming, and signal‑processing tools such as MATLAB and Simulink positions him well to contribute to algorithm development and system validation. The expected outcome is a functional, low‑complexity activity‑recognition pipeline suitable for future extensions to healthcare, gait analysis, and smart‑environment applications.

Project Title: Grid Forming converters in Wind Energy Conversion Systems

Supervisor(s): Rukmi Dutta

Email: rukmi.dutta@unsw.edu.au

Description: This project will investigate grid-forming converter control strategies for wind energy conversion systems using MATLAB–Simulink–based simulations. Students will model and analyze the dynamic behavior of wind turbine converters under grid-forming operation, focusing on stability, frequency support, and interaction with weak grids. The project is suitable for undergraduate or master’s honours level, with emphasis on power electronics, control, and power system integration.

Project Title: Grid-Forming Control of Wind Energy Conversion Systems Using MATLAB–Simulink

Supervisor(s): Clay Chu

Email: g.chu@unsw.edu.au

Description: This project will investigate grid-forming converter control strategies for wind energy conversion systems using MATLAB–Simulink–based simulations. Students will model and analyze the dynamic behavior of wind turbine converters under grid-forming operation, focusing on stability, frequency support, and interaction with weak grids. The project is suitable for undergraduate or master’s honours level, with emphasis on power electronics, control, and power system integration.

Project Title: Harnessing Imaging Rate and Throughput in Nanoworld: Novel Nanotips and Advanced Nanocantilevers Probes

Supervisor(s): Aron Michael

Email: a.michael@unsw.edu.au

Description: Manipulating, interrogating, and studying structures and dynamics at the atomic scale are essential for discovering new scientific knowledge and driving innovation across many disciplines in science and engineering. Atomic Force Microscopy (AFM) provides a powerful platform for performing such operations. However, despite substantial progress since its introduction in 1986, AFM still faces significant limitations in operational speed and throughput—particularly when imaging soft biological and macromolecular samples such as proteins and cells.

These constraints have become a major bottleneck for uncovering nanoscale dynamic mechanisms and for scaling up the use of scanning probe techniques. Addressing the limitations in imaging rate and throughput is therefore critical and would have far reaching impact across multiple scientific and technological fields. The key lies in developing an advanced active cantilever probe that: (i) achieves high resonance frequency while maintaining a low spring constant; (ii) can be integrated with ultra–high-aspect-ratio nano tips in parallel; and (iii) incorporates embedded sensing and actuation functionalities. This project aims to develop such active cantilever probes - the smallest, the fastest, and the most advanced - to significantly harness the imaging speed and throughput of AFM systems.

Project Title: Noise Analysis and Optimal Integration Design for Single-Shot SET-Based Spin-Qubit Readout

Supervisor(s): Arup George

Email: arup.george@unsw.edu.au

Description: This three-month research project investigates the fundamental noise limits of single-shot spin-qubit readout using a single-electron transistor (SET), gated charge integrator, and comparator-based digital decision. The study will develop a complete analytical noise model, quantify how integration time, capacitance, and circuit noise determine readout fidelity, and identify optimal design regimes for high-accuracy quantum measurement. Combining theory with numerical simulation, the project aims to produce practical design guidelines for next-generation semiconductor spin-qubit sensors and cryogenic readout electronics, with outcomes suitable for conference or journal publication.

Project Title: Study of post draw thermal procssing on ative optical fibres

Supervisor(s): Gang-Ding Peng

Email: G.Peng@unsw.edu.au

Description: This project is aimed at studying effects of post draw thermal procssing on ative optical fibres such as bismuth/erbium and ytterbium/erbium codoped optical fibres. This project needs a student interested in experimental photonics with a strong academic background in fibre optics and optical spectroscopy. 

Project Title: Wavelength-selective optoelectronic memristor for next-generation neuromorphic computing systems

Supervisor(s): Shimul Kanti Nath

Email: shimul_kanti.nath@unsw.edu.au

Description: Are you interested in building the next generation of brain-inspired computing hardware? This research project focuses on the design of a wavelength-selective optoelectronic memristor for neuromorphic computing, with an emphasis on implementing artificial neurons. Using Python-based simulations, you will design a Bragg filter that selectively transmits specific wavelengths of light into a broadband light-sensitive memristor device.

By controlling the optical input spectrum, we can engineer neuron-like activation behavior directly at the device level. This approach opens exciting possibilities for hardware-based neural networks applied to pattern recognition, image processing, and edge computing systems. This project is ideal for students interested in optoelectronics, computational modeling, AI hardware, and next-generation computing architectures.

interested in experimental photonics with a strong academic background in fibre optics and optical spectroscopy. 

Project Title: Biologically Inspired Speech Modelling

Supervisor(s): A/Prof. Vidhyasaharan Sethu

Email: v.sethu@unsw.edu.au

Description: Despite significant advances in speech-based AI systems, their capabilities are still no match for human abilities. For instance, most people will have no trouble having a conversation in a noisy restaurant but wouldn’t expect the speech input on their smartphone to work in the same environment. We do not have a full understanding of how the human auditory system achieves this level of performance, but we have hints.

The human brain makes use of prior knowledge of the structure of speech sounds to help decode it even in situations when the signal to noise ratio is low. The aim of this project is the preliminary investigation and development of biologically inspired speech analyses systems that can track information from speech under noisy conditions. If successful, this could be the first step in the path leading to the next generation of speech-based AI systems.

Project Title: Rethinking AI: Moving from Deep Learning to Assembly Calculus

Supervisor(s): A/Prof. Vidhyasaharan Sethu

Email: v.sethu@unsw.edu.au

Description: Despite the rapid advances in generative AI, they are no match for biological intelligence. Current AI systems depend on vast amounts of data and often fail to generalise when confronted with unseen or novel tasks, due to the lack of a world model. This project aims to develop a new paradigm for a neurosymbolic AI that moves away from conventional deep learning by unifying model structures, learning paradigms, and world models. Central to the proposed approach is a biologically inspired cognitive architecture based on Assembly Calculus. The aim of this research is to develop scalable and efficient AI capable of robust generalisation bring us closer to human-like cognition. Suggested Reading: https://www.pnas.org/doi/abs/10.1073/pnas.2001893117 https://arxiv.org/abs/2406.07715

Project Title: Bio-enhanced geomaterial hydrogen production and CO2 mineralisation

Supervisor(s): Hamid Roshan; Alex Hashemi

Email: h.roshan@unsw.edu.au

Description: By harnessing microbially-assisted chemical reaction within natural geomaterials, this process enables gold hydrogen generation while simultaneously locking away CO₂ as stable mineral phases. This dual-function process not only addresses two of the most urgent climate challenges including clean energy production and permanent carbon sequestration but also does so using abundant materials and potentially in situ methods, reducing energy input and environmental footprint. As such, it represents a transformative pathway toward scalable, sustainable, and circular climate solutions.

Project Title: Enhancing Hard Rock Excavation Efficiency Using High‑Power Microwave Rock Weakening

Supervisor(s): Hamed Lamei Ramandi

Email: h.lameiramandi@unsw.edu.au

Description: UNSW’s MERE Laboratory has recently obtaind a high‑power microwave system designed specifically for advanced rock‑mechanical testing. This project investigates the application of microwave energy to pre‑condition and weaken hard rock formations to improve the efficiency of excavation processes. By inducing thermal stresses, microcracking, and strength reduction within target rock masses, microwave treatment has the potential to significantly reduce cutting forces and tool wear in both tunnelling (including TBM operations) and mining environments. The project will quantify the extent of microwave‑induced weakening across various hard rock types, develop predictive models for excavation performance improvement, and assess the operational feasibility of integrating microwave-assisted rock breaking into modern excavation systems.

Project Title: Satellite-based Remote Sensing of Coal Mine Methane Emissions

Supervisor(s): Simit Raval

Email: simit@unsw.edu.au

Description: Recent advances in satellite remote sensing now allow methane plumes and hotspots to be detected from space using both broad-coverage missions (e.g., Sentinel-2/3, Landsat) and dedicated methane missions (e.g., EMIT, GHGSat, Carbon Mapper). This project will use an integrated multi-satellite approach to observe methane emissions from coal mines with an aim to improve emission quantification by comparing multiple retrieval and inversion pathways and explicitly evaluating uncertainty sources. This opportunity is suitable to a master’s students who has demonstrated capabilities in remote sensing of air quality/atmospheric chemistry with a focus on greenhouse gases emissions. 

Project Title: Global hail risks on PV

Supervisor(s): Dr Abhnil Prasad

Email: abhnil.prasad@unsw.edu.au

Description: This project develops the first global, data‑driven assessment of hail risk to photovoltaic (PV) systems by integrating weather reanalysis and climate model outputs with PV vulnerability thresholds. Using high‑resolution meteorological datasets and CMIP6 climate projections, the study quantifies global patterns of hail occurrence, intensity, and future changes under multiple emissions scenarios. These hazard metrics are combined with engineering-based PV damage thresholds to construct a geospatial hail–PV risk index.

The resulting global maps identify current and emerging hotspots of hail-related PV vulnerability, providing insights relevant to system design, asset management, and climate‑resilient energy planning. Outcomes support insurers, developers, and policymakers in anticipating hail-related losses and guiding robust PV deployment in a changing climate.

Project Title: Contactless measurements of solar cells and modules

Supervisor(s): Ziv Hameiri

Email: ziv.hameiri@unsw.edu.au

Description: Development of contactless electrical measurements for solar cells and modules using luminescence imaging and machine learning models.

Project Title: Machine learning applications for improved contactless series resistance imaging

Superviser(s): Zubair Abdullah-Vetter, Yan Zhu, Ziv Hameiri

Email: z.abdullahvetter@unsw.edu.au, yan.zhu@unsw.edu.au, z.hameiri@unsw.edu.au

Description: The photovoltaic industry is growing rapidly and is projected to supply about 80% of global renewable power by 2030. To support this expansion, reliable and fast characterisation techniques are required for solar cells in mass production lines. A key factor in improving solar cell efficiency is optimising its electrical parameters, particularly the series resistance (Rs). The most common way to determine (global) Rs of a solar cell is by current-voltage (I-V) measurements, which do not provide spatial insights. While electroluminescence (EL) imaging approaches exist to close this gap, reliable contact can be challenging or incompatible with advanced designs such as multi-busbar (multi-BB), zero-busbar (0BB), or back-contact (BC) cells.

The ACDC group has developed methods to obtain Rs images of solar cells using contactless characterisation approaches. The approach is mostly consistent with contacted Rs measurements, and they are relatively indicative of regions of high series resistance. However, due to the imaging approach, artefacts and recombination features may impact the quantitative accuracy of the measurements. This will affect both spatial and global value accuracies. To overcome these image artefacts, we propose using deep learning approaches.

By training the models to identify the (often small) artefact features and adjust them to match or be closer to the contacted Rs image (or global value), we can reduce the gap between contacted and contactless Rs imaging approaches. This will enable the use of contactless Rs imaging to directly characterise and optimise the electrical performance of solar cells, contributing to the ever-increasing throughput of mass production lines and enabling the consistent production of high-efficiency solar cells.

Project Title: Embedding sustainability-related competencies in engineering curriculum

Supervisor(s): Santosh Shrestha, Gavin Conibeer

Email: s.shrestha@unsw.edu.au, g.conibeer@unsw.edu.au

Description: This project will explore practical ways to strengthen sustainability learning within undergraduate and postgraduate engineering programs. Students will review the current curriculum, identify gaps in sustainability-related competencies, and develop short, hands‑on micro-project activities to address these gaps. The project will deliver a set of teaching modules that can be integrated into individual courses or applied more broadly across an engineering degree.

Project Title: Dynamic life cycle assessment of utility-scale photovoltaic systems

Supervisor(s): Rama Sharma and Prof. Ziv Hameiri

Email: rama.sharma@unsw.edu.au, z.hameiri@unsw.edu.au

Description: Solar photovoltaics (PV) are foundational to the global clean energy transition. By mid-century, installed global PV capacity is projected to exceed ten terawatts, delivering a dominant share of renewable electricity and driving deep reductions in greenhouse gas emissions. Although PV is often called “zero emissions” during operation, this label obscures substantial environmental impacts arising from raw material extraction, module manufacturing, transportation, component replacement, and end of life (EoL) processes. Life Cycle Assessment (LCA) has become the standard tool to evaluate such cradle to grave impacts, commonly using indicators such as Global Warming Potential (GWP), Energy Payback Time (EPBT), and resource depletion.

For utility scale PV systems, reported GWP values range from 10 to 90 g CO₂ eq/kWh and EPBT from 0.6 to 2.2 years, reflecting sensitivity to assumptions about module efficiency, degradation rates, energy mix in manufacturing, and disposal strategies. Yet traditional LCA methods are static - they aggregate emissions over the lifetime and assume constant performance metrics. Such an approach overlooks when emissions occur: emissions released today have greater short term warming impact than those emitted later, particularly in the context of decarbonising electricity grids or net zero climate targets. Dynamic Life Cycle Assessment (dLCA) overcomes this limitation by explicitly incorporating the temporal dimension. Rather than aggregating emissions, dLCA tracks their timing and evaluates their climate impact using time sensitive metrics like Absolute Global Warming Potential (AGWP) and Absolute Global Temperature change Potential (AGTP).

These metrics elucidate how emissions contribute to radiative forcing and temperature change at relevant policy horizons (e.g., 2030, 2050), enabling more policy aligned environmental assessments. Despite its potential, dLCA remains underutilised in the PV sector and has never been systematically applied at utility scale, where the majority of global PV capacity is deployed. This project addresses that gap by developing and applying a dLCA framework for utility scale PV systems, integrating non-linear degradation modelling, component replacement dynamics, evolving grid carbon intensity, and realistic EoL scenarios. The result will be a novel, robust framework capable of mapping time dependent climate impacts of large-scale solar energy and enhancing decision makers’ understanding of PV’s role in the decarbonisation pathway. 

Project Title: Advanced characterisation method for investigating silicon surface recombination

Supervisor(s): Huy Tuan Anh Le

Email: huytuananh.le@unsw.edu.au

Description: Most modern silicon solar cells rely on their outstanding surface passivation to achieve very-high efficiencies. The current trend of designing thinner solar cells for cost reduction and enhanced open-circuit voltage, coupled with recent advancements in the bulk quality of silicon wafers, further underscores the sensitivity of modern cells' performance to surface recombination. Consequently, minimising this loss mechanism becomes exceptionally crucial. To mitigate surface recombination, reducing the density of interfacial defects and/or controlling the carrier populations near the silicon surface are key factors.

The former can be achieved by saturating the dangling bonds with an additional dielectric layer or chemical species such as hydrogen. Meanwhile, the latter can be accomplished through methods such as doping, introducing fixed charges in the dielectric films, or controlling the materials’ work function. All these approaches aim to amplify the imbalance between the concentrations of electrons and holes near the silicon surface. Therefore, gaining a deeper understanding of surface carrier population control is critically important for further advancements in surface passivation and carrier selectivity.

It would be even more beneficial if this understanding could be achieved under field operating conditions and for metallised structures, which common lifetime measurement techniques cannot address. The main project aims are to: • Develop advanced characterisation methods to investigate silicon surface recombination in both metallised and non-metallised structures under field operating conditions through surface carrier population control. • Establish models to extract critical parameters, such as temperature-dependent carrier capture cross-sections, the density of interfacial defects, bulk lifetime, fixed charges, and more. • Save the world!"

Project Title: Machine Learning Discovery of Luminescent Nanomaterials for High-Efficiency and Durable Silicon Solar Modules

Supervisor(s): Mahesh Suryawanshi

Email: m.suryawanshi@unsw.edu.au

Description: This project will use machine learning and data-driven approaches to accelerate the discovery and optimization of luminescent nanomaterials for next-generation photovoltaic (PV) encapsulants. The focus is on identifying quantum dot (QD) compositions and polymer–nanomaterial combinations that maximise photoluminescence quantum yield (PLQY), spectral overlap with silicon absorption, and long-term UV stability. The student will work at the interface of materials science and artificial intelligence, developing predictive models using literature data and experimental datasets to correlate composition, structure, and optical performance. The project will involve data curation, feature engineering, regression modelling, and performance benchmarking against known luminescent systems, such as perovskite and copper-chalcogenide nanocrystals.

Where appropriate, computational predictions may be validated through preparation and optical characterisation of selected QD–polymer films (UV–Vis spectroscopy, steady-state photoluminescence). The expected outcomes include a ranked materials design map and the identification of promising luminescent systems suitable for improving the efficiency and durability of silicon solar modules. This project provides training in advanced materials informatics, renewable energy technologies, and experimental validation within UNSW’s School of Photovoltaic and Renewable Energy Engineering.

Project Title: In-depth analysis of photoluminescence mappings for perovskite solar cells, assisted by machine learning techniques

Supervisor(s): Arthur Julien

Email: a.julien@unsw.edu.au

Description: Perovskite solar cells are promising candidates for low-carbon electricity production due to their affordability, ease of fabrication, and low resource requirements. However, stability issues must be addressed. Photoluminescence measurements are essential as they provide rapid and non-destructive insights into semiconductor quality. Moreover, hyperspectral acquisitions offer spectrally and spatially resolved PL data, typically analysed using physics-oriented approaches.

However, machine learning tools, such as convolutional neural networks, can enhance these analyses by identifying spatial features with specific behaviours or properties. This project aims at developing hyperspectral acquisitions and coupled machine learning tools to improve the understanding of ageing phenomena in perovskite solar cells.