How Uber Optimizes Code at Scale with Grafana Cloud Continuous Profiling & Grafana Assistant

Grafana

Nov 2, 2026

How does Uber use continuous profiling to improve performance and resource efficiency across several million CPU cores? In this Grafana Cloud case study, Senior Researcher Chris Zhang from Uber’s Frontier Tech Labs explains how the team uses Grafana Cloud Profiles and Grafana Assistant to identify expensive code and scale code optimization across its production systems.

Uber’s microservice environment spans multiple programming languages, reliability tiers, and stateful and stateless services. See how the team deployed cost-effective, always-on production profiling to give hundreds of engineers code-level visibility into resource usage.

Learn how Uber:

Deployed continuous profiling across a complex microservices environment.
Made profiles accessible to hundreds of engineers to identify expensive code.
Used code-level performance insights to improve resource utilization and help prevent regressions.
Integrated Grafana Assistant into engineering pipelines.
Combined generative AI with neurosymbolic analyses to continuously optimize code with little human oversight.

Chris shares how Uber is moving from continuous profiling to continuous optimization, making code optimization more accessible while improving cost-effectiveness and customer experience.

Speaker:
Chris Zhang, Senior Researcher, Frontier Tech Labs, Uber

Chapters:

00:00 Introduction and Goals

01:45 Why Profiling Matters

05:27 The Problem with Random Sampling

10:15 Topology-Aware Profile Assignment

12:08 Pyroscope Proxy and Kill Switch

16:02 CPU and Mutex Fixes

20:42 Grafana Assistant and Agents

23:28 Neurosymbolic Optimization

26:06 Key Takeaways

Links/resources:
Explore ObservabilityCON: https://grafana.com/events/observabilitycon/
Join us at GrafanaCON 2027: https://grafana.com/events/grafanacon/
Grafana Labs success stories and case studies: https://grafana.com/success/
Other talks from ObservabilityCON 2026: https://www.youtube.com/playlist
Get started with the Grafana Cloud forever-free tier: https://grafana.com/g/cloud