43Build a streaming aggregation with exactly-once output, bounded state, and a defined policy for events that arrive an hour late.▼hardDatabricksSnowflakePalantir1 replies◆ premiumStreaming interviews separate people who have run a job at 3am from people who have read the docs. The hard parts are not the windowed aggregate, they are bounding state so the job does not die, and choosing what a late event does to an already-emitted result.Open full answer →
49Orchestrate a multi-agent system over a 500-page merger agreement to produce a redlined risk summary that never contradicts the source.▼expertHarveyHebbiaPalantir1 replies◆ premiumFive LLM agents over a hundred steps, and the bar is zero contradictions with a 500-page contract. The real test is the state and grounding layer, not the agent prompts.Open full answer →