LinkedIn Staff Software Engineer
May 2022 – Present - Served as technical lead for LinkedIn's Observability UX platform for two years, setting technical direction around debt reduction and testing standards, and mentoring engineers by helping each define concrete goals and develop toward delivery that was impactful and safe to ship.
- Worked to build out web observability for LinkedIn's platform-wide re-architecture, a cross-pillar initiative (iOS, Android, and web) moving to a shared authoring model. Built out the OpenTelemetry instrumentation layer from zero to 100% production traffic — including the span-validation framework, web vitals coverage, and exception telemetry — enabling performance and availability monitoring at a granularity the previous architecture couldn't support.
- Replaced a vendor metrics datasource, which offered no control over roadmap or LinkedIn-specific query support, with a purpose-built Grafana datasource plugin and migrated ~5,000 production dashboards to it with no operational disruption to dependent teams.
- Contributed to early-stage evaluation work for Observe Agent, LinkedIn's AI-powered alert-triage platform, including P0 incident dataset curation and initial LLM evaluation methodology.
- Redesigned the alert-triage experience for LinkedIn's Observability platform, reducing friction for incident commanders and on-call engineers responding to active alerts and measurably improving customer satisfaction scores.
- Drove a targeted effort to modernize and harden the test infrastructure for LinkedIn's Observability tooling, migrating to Playwright and running a sustained reliability program to ensure stability as AI-assisted development accelerated the team's delivery pace.
- Built conversion tooling for a large-scale, ongoing migration of 80,000+ production dashboards from a legacy metric system to its modern MDM-based equivalent (some dashboards over a decade old and still actively used), focusing on the read path and producing metric-level equivalence between old and new systems.
- Drove the end-of-life of a 15-year-old site performance monitoring tool, establishing feature parity in its modern replacement and orchestrating the transition so dependent teams experienced no operational disruption.
- Maintained LinkedIn's open-source web benchmarking tool, TracerBench, as part of a shift-left performance initiative, delivering robust benchmarking across multiple LinkedIn web products at scale.
- Worked to define and establish performance and availability metrics across LinkedIn's Observability tooling, creating clearer visibility into the health of the platform itself.