Curriculum vitae

Neurally inspired hybrid local–distributed algorithms for GPS-free localization and mapping on heterogeneous architectures, combining local learning rules with reinforcement learning for real-time edge deployment. Matured neuromorphic and analog in-memory prototypes into a funded multi-year algorithm–hardware co-design programme. As principal investigator I direct roughly $4M across four grants — three LDRDs (2025–2028) and a DOE ASCR award — and lead cross-functional teams of four to nine.

Hardware-constrained streaming algorithms for object tracking in remote sensing, beating the state of the art while cutting energy by over 95% through analog computational elements. Hardware-aware quantisation enabling on-device xLM inference, and recurrent, graph and physics-informed networks for petascale spatiotemporal prediction.

Built a biologically grounded architecture and learning rule for temporal prediction that outperformed the state of the art on short- and long-horizon time series, with applications to lifelong and on-device learning. Also designed explainable models of egocentric–allocentric reference-frame transformation.

Hyperdimensional-computing models of working memory for symbolic and composable AI, plus the EEG experiments and causal frequency–time analyses that tested their predictions, for commercial and government clients.

Cross-lab platforms for standardised neural and behavioural analysis, now used by six or more independent groups; multimodal data pipelines and SQL databases; primary statistical analyst on time-series, GLM and dynamical-systems methods.