quality_assurance · saas · workflow

PerfInsights: GenAI-powered detection of Go performance antipatterns at Uber

Optimizing Go services at Uber required deep expertise and significant manual effort, with profiling and analysis taking days to weeks; in March 2024, the top 10 Go services alone accounted for more than multi-million dollars in compute spend, making performance tuning prohibitively expensive and non-trivial for most teams.

How it works
Common implementation structure
How this type of workflow is generally built, generalized across documented cases — not tied to any one vendor's stack. Click any stage to read what happens there. Specific products that implement these stages appear in “Tools commonly seen” below.
Stage 1 · Fleet-wide production profiling
PerfInsights collects CPU and memory profiles from production services using Uber's daily fleet-wide profiler during peak traffic periods.
Tools used
PerfInsightsLLMLLMCheckOptix
Outcome

PerfInsights reduced performance analysis from days to hours, cut engineering time per issue from 14.5 hours to almost 1 hour (93.10% savings), reduced false positives from over 80% to the low teens, and produced hundreds of merged diffs driving compute cost reductions across Uber's Go services.

What failed first

Initial single-shot LLM-based antipattern detection produced inconsistent and unreliable results—responses varied between runs, included hallucinations, and often generated non-runnable code—with false positives exceeding 80%.

Results
Time savedtasks that once required days now take hours
Volumeover 80% to the low teens
Cost replacedmore than multi-million dollars
Running sinceHackdayz 2024
Source

https://www.uber.com/en-NL/blog/perfinsights/

How we source this →

Grounding & classification
Source type: technical build writeup
52 fields verified against source quotes.
anomaly detectioncode generationmulti agent workflowquality inspectioncode diff prknowledge basebuilder submittedfailure mode describedhuman review describedmetric backednamed customerproduction runtime claimedsource backedtools describedworkflow describedlogisticssoftwareaccuracy improvementcost reductionemployee productivityerror reductiontime savedtechnical build writeupquality assuranceai draft human approvalextract classify routemonitor detect alert