Nick Berry Resume

Tech Lead for SPICE, a multi-track project to prepare LinkedIn for data center growth over 10 years, transparently scale the network, and improve fleet-wide efficiency and resource accountability. Responsible for vision, strategy, and driving execution across dozens of teams.
Designed and ran production experiments with network latency, validating the ability to extend distances between data centers up to 4x farther, avoiding the cost ($XXX millions in capex) and complexity of a greenfield build.
Drove fleet-wide network enhancements and optimizations. Unified QoS across datacenters and backbone with host-based tagging unlocking the ability to mix workload types as well as multiple second order effects. Enabled DCTCP for intra-region congestion control, cut retransmits at host and switch-level up to 90%.
Co-led Fleet Analytics, LinkedIn's first unified capacity visibility platform and dataset. Enabled teams to identify efficiency gaps, track resource usage, and decommission under utilized hosts. I aligned senior leadership on the use of the data into capacity forecasting and weekly reviews, identifying opportunities to binpack (in-place and upgrades), and make data-driven reductions in capacity forecasts up to 20%. Partnered with the Capacity Engineering team to ensure timely implementation of dynamic CPU and memory tuning across thousands of applications, resulting in increased capacity of shared pools.

SRE Lead for LinkedIn's migration to Azure. Responsible for defining and driving Operational Readiness, VM sizing recommendations for 6k applications across 250k, and developed principles for standardized VM images (stopping unbounded growth to only 10). Led effort to reduce system overhead by eliminating unnecessary agents (recovering ~1M CPU cores).
Led fleet-wide remediation of Meltdown/Spectre vulnerability, first fleet-wide OS upgrade effort. Owned assessment, communications, coordination, and execution - responsible for building initial automation and unlocking self-service for multiple black-box systems which became the foundation of future OS upgrades.
Responsible for transitioning from RHEL to CentOS with custom kernels, then aligned LinkedIn and Microsoft leadership to adopt Azure Linux as the standard with a scheduled lifecycle. Drove retirement of RHEL6 and RHEL7.
Partnered across teams to eliminate standing production access. Collaborated on the design and implementation of certificate-based, just-in-time shell access. Responsible for ramp, communication, and socializing changes across of Engineering.

Built LinkedIn's Infrastructure SRE organization from the ground up. As one of LinkedIn's founding SRE managers, I helped build the model that became LinkedIn's standard for reliability engagement. I scaled from a small team to 40+ engineers (SRE and SWE) across SF, NYC, and India. Hired and developed senior management and technical leads.
Scaled observability platforms from 10M to 2B+ metrics/minute. Owned LinkedIn's metrics, visualization, alerting, and escalation, with a focus on self-service and guardrails to minimize human error.
Deployed LinkedIn's edge infrastructure strategy. Key contributor to building, deploying, and supporting Apache Traffic Server, HAProxy, and IPVS as the traffic routing tier.
Authored LinkedIn's first Python-native Kafka producer library, enabling non-Java solutions to integrate with the metrics pipeline for the first time.

Lead SE for Yahoo! Frontpage. Managed multi-tier, auto-failover stack with GSLB configuration, load distribution, and monitoring for one of the highest-traffic web properties in the world.
Redesigned release process, reducing deployment time from 2.5 hours to 20 minutes. Reduced emergency failover time from over 1 hour to 10 seconds by implementing a static failover solution with streamlined execution. Defined monitoring standards and drove implementation of application-level metric aggregation, reporting, and alerting across multiple organizations.

Nick Berry