Build and evaluate retrieval and answering systems with evidence-linked questions grounded in extreme heat emergency plans.

Explore watermark progression, state evolution, and late-event impacts for temporal join reasoning and stream engine validation.

Turn written modeling requirements into block model scripts with examples covering parameter derivation and operation workflows.

Train models to determine FTTB provisioning eligibility from coverage, resources, port status, and service rules.

Pair continuous operating telemetry with annotated state changes to evaluate next-stage state prediction for sodium-ion energy storage systems.

Turn virtual experiment descriptions into structured records of measurements, observations, and conditions.

Use Kubernetes workload manifests and validation outcomes to help models identify security risks in cloud-native deployment configurations.

Evaluate whether models can detect ambiguity in question bank text, judge answerability, and explain their conclusions using supporting evidence.

Analyze requests, responses, objections, and disputes in commercial litigation discovery documents to evaluate legal document understanding.

Train models to answer questions about building fire systems using source material, while recognizing when the evidence is insufficient.

Turn road signs, lane restrictions, vehicle classes, and time conditions into queryable rule relations for Robotaxi maps.

Connect menswear fit and size queries with relevant catalog information to improve e-commerce search and shopping experiences.

Train models to reason across clock paths, synchronization, and jitter budgets in mixed-signal systems.

Evaluate cross-catalog room matching with labeled record pairs and key configuration differences.

Train models to handle sportswear product variants and pre-checkout actions accurately with tool-use traces from shopping workflows.

Explore the effects of elastic scaling decisions through aligned workload, control action, and system response trajectories.

Train models to verify balance adjustments and reconciliation results using cash ledgers, bank statements, and outstanding items.

Evaluate how well models understand commercial litigation motions, from governing standards and argument logic to requested rulings.

Improve gaming PC and component retrieval with retail catalog documents paired with queries that reflect real-world shopping needs.

Train models to spot non-compliant operations in selective lignite mining using seam constraints and step-level verification results.