The field of scientific machine learning is witnessing significant advancements in data assimilation and surrogate modeling. Researchers are exploring innovative methods to improve the accuracy and robustness of predictions in complex dynamical systems. One notable direction is the development of new data assimilation techniques that incorporate predictability-aware regularization terms, which have shown to enhance estimation performance and forecast skill. Another area of focus is the investigation of scaling laws for neural surrogates, which can help optimize dataset generation and computational resources. Furthermore, there is a growing interest in uncertainty quantification for reduced-order surrogate models, with researchers proposing post hoc, model-agnostic frameworks for predictive uncertainty quantification. Noteworthy papers in this area include: Variational Data-Consistent Assimilation, which introduces a new class of 4D-Var methods that outperform standard 4D-Var in reducing error and bias. Towards Multi-Fidelity Scaling Laws of Neural Surrogates in CFD, which provides the first study of empirical scaling laws for multi-fidelity neural surrogate datasets. In Situ Training of Implicit Neural Compressors for Scientific Simulations via Sketch-Based Regularization, which presents a novel in situ training protocol for implicit neural compressors. Uncertainty Quantification for Reduced-Order Surrogate Models Applied to Cloud Microphysics, which introduces a post hoc, model-agnostic framework for predictive uncertainty quantification in latent space ROMs. Efficient probabilistic surrogate modeling techniques for partially-observed large-scale dynamical systems, which investigates and compares various extensions to the flow matching paradigm.