Customer-obsessed science

We present ProbHardE2E, a probabilistic forecasting framework that incorporates hard operational/physical constraints, and provides uncertainty quantification. Our methodology uses a novel differentiable probabilistic projection layer (DPPL) that can be combined with a wide range of neural network architectures. DPPL allows the model to learn the system in an end-to-end manner, compared to other approaches

Is bigger always better for time series foundation models? With the question in mind, we explore an alternative to training a single, large monolithic model: build-ing a portfolio of smaller, pretrained forecasting models. By applying ensembling or model selection over these portfolios, we achieve competitive performance on large-scale benchmarks using much fewer parameters. We explore strategies for designing

Remote sensing imagery is a a rich source that serves a wide range of applications including urban planning, land management, environmental monitoring, and digital map enrichment by detecting roads and building outlines. Nevertheless, occlusions from trees often hide important features like roads from bird's-eye viewpoint. To tackle this problem, we propose GeoRoadInpaint model, leveraging stable diffusion

Therapeutic antibody discovery remains slow and resource-intensive, with traditional methods providing limited control over epitope selection. We present a workflow for de novo nanobody design applied to a novel Desmoplastic Small Round Cell Tumor target encompassing four stages: (1) epitope identification guided by our hotspot recommendation agent using physical chemistry-based structure and sequence analysis

Agentic AI systems can access vast data but struggle to apply domain expertise, namely the contextual understanding of how to use specialized information. This paper presents a practical framework for encoding such expertise, demonstrated with the National Football League (NFL) through NFL Fantasy AI, a production system delivering analyst-grade fantasy football advice, as assessed by NFL Pro analysts.