Independent researcher in Physical & Geometric AI (2023–present), working at the intersection of geometric deep learning, Bayesian and uncertainty-aware architectures, and physics-informed, curvature-aware optimization. Author of 8 first-author preprints produced in under 10 weeks of active writing and validated entirely on free-tier cloud GPUs (Kaggle/Colab T4 15GB, P100 16GB), demonstrating an ability to identify, formalize, and complete original theoretical contributions independently and at pace. Current work extends this program toward embodied prediction and world models — Dynamic LeJEPA (Maximum-Entropy Representations for Sequential Prediction), building on Balestriero & LeCun's LeJEPA framework — alongside a second line of work on graph-grounded safe routing for adaptive world models. Seeking a doctoral research position to develop this program under formal supervision, with 8+ years of applied AI/ML/CV/GenAI leadership providing the engineering depth to design, implement, and rigorously validate large-scale experiments.