Oliver Gould

Systems Engineer · Technical Leader · Creator of Linkerd

San Francisco, California
[email protected] · olix0r.net · GitHub · LinkedIn

Profile

Systems engineer with 20 years building distributed infrastructure; co-founder and former CTO of Buoyant. Focused on the layer that decides how workloads find, trust, and talk to each other: workload identity and mutual TLS by default, a memory-safe Rust data plane, and verifiable release supply chains. Now exploring how those ideas carry over to agentic software.

Experience

Buoyant — Co-founder & CTO

2015–April 2026

Co-founded Buoyant and created Linkerd; led technical strategy, architecture, and engineering, while staying hands-on, at a globally distributed company of about 25 people.

  • Created Linkerd. Saw that containerized, polyglot microservices would need the networking capabilities of frameworks like Finagle without requiring application rewrites; defined and built the original service-mesh approach.
  • Led the Linkerd 2 pivot. Drew on early production Kubernetes deployments to recognize the limitations of the JVM/DaemonSet architecture; drove the transition to a lightweight, purpose-built Rust sidecar, putting a memory-safe language in the path of production traffic, and communicated the new direction through major KubeCon talks and live demos.
  • Designed and built Linkerd's identity, mutual TLS, and authorization policy. Each meshed workload's identity is bound to its Kubernetes ServiceAccount, and service-to-service traffic gets mutual TLS automatically, with no application changes. Workload certificates live under 24 hours and rotate automatically; the cluster issuer is replaceable, so operators can rotate it or plug in their own CA. Added an ALPN-negotiated proxy-to-proxy transport carrying identity and routing metadata, and the policy layer controlling which workloads may talk to which.
  • Developed Rust networking infrastructure. Hired and worked with maintainers of Tokio, Hyper, and related foundations; worked hands-on in the Linkerd proxy, applying Finagle-inspired Service, middleware, load-balancing, and parameterized stack abstractions.
  • Designed adaptive zone-aware load balancing (HAZL). Built an endpoint-scaling, locality-aware approach to reducing AWS cross-AZ costs while preserving failover and availability; required coordinated behavior across the proxy, control plane, and multi-cluster deployments.
  • Built enterprise release and supply-chain tooling. Implemented the OCI-based packaging system for Buoyant enterprise releases, distributing SLSA provenance and SPDX SBOMs using OCI 1.1 referrers to enable verification and policy gating.
  • Guided Linkerd from CNCF acceptance (2017) to graduation (2021), the first service mesh to graduate.
  • Engineering culture. Established a problem-first RFC process; mentored engineers; handled customer architecture reviews and sensitive production incidents.
  • Served on Buoyant's board of directors alongside Benchmark Capital, contributing to company strategy and fundraising.

Twitter — Senior, then Staff Software Engineer

2010–2015

Tech lead for Twitter's Observability and then Traffic infrastructure as it evolved into a service-oriented architecture of hundreds of microservices across multiple datacenters. Worked closely with the Core Systems Libraries team responsible for Finagle and related distributed-systems infrastructure.

  • Observability. Tech lead for the initial production deployment of a centralized metrics platform, ingesting roughly 170 million metrics per minute from tens of thousands of endpoints (2013), that became Twitter's critical company-wide on-call monitoring system.
  • Hands-on development. Wrote the original metrics collector and created Colony, a query language connecting observability to service discovery.
  • Traffic and routing. Tech lead for the initial production deployment of Finagle Dtabs for dynamic, contextual routing; service networking infrastructure came to support every production service, including frontends. Collaborated on discovery, RPC, and load-balancing integration.

Yahoo! — Senior Systems Architect

2007–2010

Stevens Institute of Technology — Systems Administrator

2005–2007

Current Work

Independent — exploring agent infrastructure

April 2026–present

Hands-on exploration of how to build with agents, focused on two questions:

  • WebAssembly as a capability boundary for agent harnesses. Can Wasm's deny-by-default sandbox and explicit imports give a harness a small, auditable surface for what an agent may do?
  • atproto for agent state. Can a PDS serve as a durable, signed, publicly readable store, with typed records and reconciliation-style reads akin to Kubernetes resources?

Selected Technologies & Expertise

  • Systems: Distributed systems, networking, RPC, load balancing, service discovery, observability, reliability.
  • Security: Workload identity, mutual TLS, authorization policy, software supply chain (SLSA, SBOMs), memory-safe systems.
  • Languages & platforms: Rust, Go, Kubernetes, Linux, WebAssembly.

Education

Stevens Institute of Technology — B.S., Computer Science, 2006

Google Summer of Code (2007) — Worked on porting ZFS to NetBSD: kernel integration and volume management. Details