Projects
Some things I’ve done
SelfTherapy
I co-founded and am CTO of SelfTherapy PTY LTD. I’ve designed, delivered, and maintained an entire AI-assisted therapeutic platform under that brand. I architected and implemented the complete system across backend, multi-platform mobile applications, and web interfaces.
SelfTherapy was designed to target the enormous gap between mental health needs and available support, while providing a safe haven for people to carefully explore and unpack their experiences. While we acknowledge that in-person mental health engagement remains the gold standard, having an on-hand, tightly optimised mental health AI chat experience allows far more people to access assistance when needed. We were, and still are, concerned with how default LMM clients are managing this matter, and wanted to provide a solution to this problem, while leveraging AI to do what couldn’t be done before.
The SelfTherapy Core backend is a Django DRF system, integrated with custom middleware that calls to Firebase for authentication, along with complex model structures that allow organisation of personalised information inherited from the AI interactions. It is deployed on Heroku infrastructure. I also designed an entire beautiful and fully functional frontend experience in Flutter with unique chat features that assist users to identify emotions, thoughts and sensations during their engagement with the AI assistant.
In the background, AI agent layers interweave with gated database access in order to provide a coherent and safe storage and retrieval of contextual user information. The system has working integration with RevenueCat for subscription service support, with automated webhook updates to user models. The system persistently assesses for multi-domain risk, and recommends the user for assistance escalation and off-boarding in a therapeutically cautious way by more than one mechanism. I developed essentially all features, from basic account management to advanced AI therapeutic controls in order to craft a useful and safe experience.
During development, I generated blinded AI-powered stress test conversation simulations that were used for tuning and risk identification.
I have also built a companion app under SelfTherapy branding called SelfTherapy EMDR that assists users through the EMDR protocol: one of the most effective forms of trauma processing known to science. The system proved astoundingly effective in testing, but is currently only available for research trial requests due to the inherently high-risk nature of independent traumatic content processing.
SelfTherapy is currently available for open use.
You can try the app here.
MindCompass
In collaboration with Anywise, I led technical development of MindCompass, a shared-codebase adaptation of the SelfTherapy platform optimised for mental health support in military and first-responder contexts, a target demographic with unique needs and well documented mental health challenges.
SelfTherapy was expanded to handle dual deployments to multiple, uniquely branded frontend versions with the ability to toggle context specific features. I built the entire structure to handle this version toggling for both the frontend and backend, also expanding capabilities for A-B testing.
The system was highly praised when demonstrated at Land Forces 2024.
The MindCompass system is currently under review for medical research allocation in collaboration with Anywise, Gallipoli Research and Soldier On.
The app is currently available for testing.
You can try the app here.
FABHUMS
In collaboration with Anywise, I contracted for frontend Vue.js development during crunch-time on the FABHUMS project, a bridge health monitoring system developed for and deployed to military clients.
FABHUMS is a real-world system, blending software and hardware interfaces, designed for enhancing reliability, efficiency, safety, and ensuring operational integrity, all while reducing lifecycle costs for critical infrastructure.
You can learn more about FABHUMS here.
Biophysical Research
During my PHD (Paused), I designed and conducted computational biophysics research, investigating immune system protein dynamics using national supercomputing infrastructure (NCI) for large-scale simulation and analysis.
I’ve spent years developing Python automation pipelines producing and processing terabytes of complex data across distributed GPU/CPU HPC systems.
I’ve researched ion-channel proteins and the infamous SARS-CoV-2 spike protein, but my primary interests became immunological, focusing on the proteins TLR7 and DECTIN1. To do this, I used high-throughput FEP calculations during molecular dynamic simulations of carefully prepared digitally reconstructed protein structures, where hundreds of atomic-scale simulations would be run simultaneously on GPU supercomputer clusters in order to reconstruct the free energy impact of point-mutations, providing atomic-scale answers to real-world disease expressions such as specific kinds of Lupus.
My primary simulation engine is Amber, though I’m also adept at NAMD and have some experience running GROMACS. For visuals, I like Pymol for its beautiful, customisable rendering engine, and VMD for its fast visualisation of million+ atom system trajectories.
My research has been published in tier-1 journals Nature and Science Advances.
Check out my ResearchGate to find links and summaries of the papers I have been involved in.
You can read one of our press releases for our Nature paper here.