Hud deploys a zero-config production sensor that captures live service and function-level telemetry from deployed code, with a focus on AI-generated software. The tech stack (AWS, GCP, Azure, PostgreSQL, ClickHouse, Redshift, BigQuery across compute and storage) reflects infrastructure-monitoring DNA; the active projects reveal an emerging focus on evaluation datasets and auto-remediation prompts, signaling a pivot toward AI-driven incident investigation. Early hiring is balanced between engineering and sales, suggesting founder-led initial sales cycles are giving way to structured GTM.
Hud builds a runtime code sensor designed to detect and fix issues in AI-generated and AI-assisted code before they reach production. The product installs with zero configuration and automatically ingests live data from running services, capturing the context needed to debug failures in agentic workflows. The company operates from New York and is actively hiring engineers and sales roles across the US and Israel, with a focus on solving the intersection of AI reliability and production observability at scale.
Hud runs on AWS, GCP, and Azure for compute and cloud services. Backend uses Node.js, Python, Go, and Java; data pipelines leverage PostgreSQL, ClickHouse, BigQuery, and Redshift for storage and analytics; orchestration runs on Docker and Kubernetes.
Core projects include the runtime code sensor, distributed-systems tracing, low-overhead instrumentation, data pipelines, evaluation datasets for AI workflows, and auto-remediation prompt optimization for incident investigation.
Other companies in the same industry, closest in size