The field of robotics and AI is witnessing significant developments in research infrastructure, with a focus on improving reproducibility, scalability, and collaboration. Recent innovations have led to the creation of unified package-management frameworks, open-source software and mechatronics infrastructures, and novel approaches to state estimation and control. These advancements are enabling researchers to develop and deploy custom robotic systems more efficiently, and are accelerating progress in areas such as wearable robotics and autonomous systems. Notable papers in this area include: Pixi, which presents a unified package-management framework for robotics and AI, ensuring bit-for-bit reproducibility across platforms. Epically Powerful, an open-source robotics infrastructure that streamlines the development of wearable robotic systems. N-ReLU, a zero-mean stochastic extension of ReLU that replaces negative activations with Gaussian noise, enhancing optimization robustness. CENIC, a continuous-time error-controlled integrator that brings together recent advances in convex time-stepping and error-controlled integration, providing guarantees on accuracy and convergence. Discovering and exploiting active sensing motifs for estimation, which introduces a framework to refine sporadic estimates from bouts of active sensing, combining data-driven state and observability estimation with model-based estimation.