Uber
1 Tool · 30 Intel
Uber Technologies is a NYSE-listed ride-hailing and mobility company founded in March 2009 by Travis Kalanick and Garrett Camp. Originally launched as UberCab, it pioneered on-demand ride-hailing and has since expanded into food delivery, freight logistics, and micromobility, operating in roughly 70 countries.
Is this you? Sign in with X to claim this profile.
Tools
Security stack for AI agents pairing a telemetry sensor, a 300+ task benchmark, and a dual-agent detector.
Intel
Uber's DragonCrawl uses LLMs to read mobile app screens and decide what to tap the way a human tester would, adapting to UI changes without script maintenance across 50+ languages, cutting the test upkeep that had consumed 30-40% of engineers' time.
Uber extends Kubernetes with hierarchical resource pools for its Ray based ML jobs. Each team gets a reservation, idle capacity is borrowed elastically across pools, and preemption returns it when the owner needs it again, with max-min fairness driving the entitlement math.
Uber engineering details uReview, an AI code-review system whose prompt-chaining pipeline (comment generation, filtering, validation, deduplication) reviews over 90% of weekly diffs, with 75% of its comments marked useful by engineers.
After making thousands of microservices MCP-callable, Uber found agent actions traceable only to a generic service identity, such as a PR opened by a Monitoring Agent with no link to the on-call engineer behind it. This covers the identity and delegation model built to fix attribution.
Uber engineering describes PerfInsights, a system that combines fleet-wide CPU/memory profiles with GenAI static analysis, validated by an LLM jury, to find and fix Go performance antipatterns, cutting optimization work from days to hours.
Uber built a File Semantic Analyzer that uses LLMs to classify and summarize files leaving its environment, replacing keyword-and-regex DLP that drowns reviewers in false positives. Covers the classification pipeline and how semantic labeling cuts manual oversight of data egress.