News
RESEARCH / MILESTONES / IMPACT
Recent work and selected milestones. The five newest items are highlighted; the complete archive follows in the same format.
Latest
A causal perspective on world models
Our perspective with Fabrizio Russo connects observations, representations and causal structure; the manuscript reports acceptance at the UAI 2026 Causality in Decision Making workshop. Paper and status
Blurgs AI raises $2.2 million
Funding led by Pravega Ventures and Shastra VC supports the next stage of our maritime intelligence work. Business Standard
Steering diffusion representation learning
A preprint with Rajat Rasal, Tian Xia and Ben Glocker examines how optimisation trajectories shape learned representations. Preprint
UniBrain: understanding incomplete MRI inputs
The UniBrain manuscript is online, with MICCAI 2026 acceptance reported in the authors' arXiv record. Paper and status
Global Ocean Innovation Challenge pilot selection
TNC and Newlab selected Blurgs.AI for an electronic-monitoring training and analytics pilot; this records the selection milestone rather than a measured conservation outcome. TNC announcement
Archive
Factored Classifier-Free Guidance
The updated manuscript reports acceptance to ICML 2026 and studies attribute-wise control for counterfactual image generation. Paper and status
PhD completed; EPSRC prize fellowship
The Safe & Trusted AI CDT featured my doctoral research, successful viva and award of an EPSRC PhD prize fellowship. Research interview
Object representations, counterfactuals and causal discovery at ICML
Three papers in the ICML 2025 proceedings address identifiable object representations, diffusion counterfactuals and Bayesian causal discovery. Research highlights
Explaining image classifiers through argumentation
Free Argumentative Exchanges for Explaining Image Classifiers appears in the AAMAS 2025 proceedings. Institutional record
Grounding and identifying object-centric representations
Our ICLR and NeurIPS work studies grounded representations and probabilistic slot attention. Institutional record
Three new preprints
Preprints on vector quantisation for robust segmentation, GLANCE explanations, and hierarchical symbolic reasoning were posted online. Scholar
Vector quantisation work accepted at MICCAI 2022
Work on robust segmentation with vector quantisation was accepted at MICCAI 2022. Scholar
Robustness and quantisation manuscript
Submitted work on model robustness and vector quantisation to MICCAI. Scholar
Joined the Safe & Trusted AI CDT PhD programme at Imperial
Started doctoral research at Imperial College London. CDT
Whole-slide image analysis paper
A generalized deep-learning framework for whole-slide image segmentation and analysis was published in Scientific Reports. Paper
AAAI workshop best paper award
Our work on CNN interpretability received the overall best paper award at the W3PHIAI workshop. Paper
AAAI workshop spotlight
Our work on CNN interpretability was selected for a spotlight presentation at W3PHIAI. Paper
Symbolic Pursuit extended to classification
Extended the symbolic-regression work to classification tasks and released the accompanying repository. Repository
Concept graphs for interpretability
Posted work on abstracting deep neural networks into concept graphs for concept-level interpretability. Preprint
Graduated from IIT Madras
Completed my Dual Degree programme at the Indian Institute of Technology Madras. IIT Madras
Digital pathology review
Work on segmentation and classification in digital pathology for glioma research appeared in Frontiers in Neuroscience. Paper
Brain tumour interpretability and uncertainty
Our study of interpretability and uncertainty in brain tumour segmentation networks appeared in Frontiers in Computational Neuroscience. Paper
Neural field modelling package
Released a Python package for neural field modelling. GitHub
BioExp package
Released tooling for explaining biomedical deep-learning models. GitHub
DigiPathAI package
Released deep-learning tooling for histopathology analysis. GitHub
MICCAI presentation on signet-ring cell detection
Presented the winning technique for the DigestPath signet-ring cell detection challenge. Talks
IDRiD challenge paper
The IDRiD diabetic-retinopathy segmentation and grading challenge paper appeared in Medical Image Analysis. Publications
Brain tumour segmentation package
Released deep-learning tooling for brain tumour segmentation. Repository
Visiting research internship at Stanford
Worked as a visiting research intern with the Poldrack Lab in Stanford's Department of Psychology. Lab