Customer-obsessed science

Tabular foundation models like TabPFN and TabICL achieve state-of-the-art performance through in-context learning, yet their architectures remain fundamentally opaque. We introduce KernelICL, a framework to enhance tabular foundation models with quantifiable sample-based inspectability. Building on the insight that in-context learning is akin to kernel regression, we make this mechanism explicit by replacing

Large language model (LLM) agents increasingly operate in streaming case-based reasoning (CBR) settings, where continuous improvement from past experience is crucial. Existing methods achieve this by storing past cases and retrieving similar ones as few-shot examples. This strategy fails near decision boundaries, where highly similar cases have conflicting outcomes and the discriminative factors are not

Recent agentic approaches to LLM-based kernel generation have achieved strong results on CUDA, yet emerging AI accelerators such as AWS Trainium and Inferentia remain unaddressed. Writing kernels for these chips via the Neuron Kernel Interface (NKI) is particularly challenging due to a multi-engine architecture, tile-based programming with a fixed 128-element partition dimension, and explicit memory management

Large language models can memorize information that must be removed–ranging from copyright-sensitive content (e.g., book chapters) to personally identifiable information (e.g., income)–to ensure responsible and compliant behavior. Unlearning has emerged as an efficient alternative to full retraining, aiming to remove specific knowledge. However, users may still expect model to leverage the removed information

E-commerce assistants must go beyond product search to support idea inspiration, criteria formation, comparison, and tool-grounded fact-checking over non-linear shopping journeys. Teaching these behaviors into deployable latency-constrained models is bottlenecked by post-training data: trajectories must cover the full agentic workflow with diversity and fidelity, yet desired outputs are open-ended (often