● available for select engagements

I keep very large things
from falling over.

Principal Reliability Engineer. Fourteen years on the night shift of the internet: payment rails, freight networks, and a satellite ground segment that was, on a good day, mostly working.

99.995%trailing availability, 9 years running
41 minlongest incident I owned (never again)
2.3M rpspeak traffic survived without paging me
0postmortems that blamed a person

Career

2023 — now

Principal Reliability Engineer · Orbitline Systems

Own reliability for the ground segment behind a 400-satellite constellation. Yes, the pager works in space too; it just has more latency.

  • Replaced a hand-rolled failover with a quorum-based control plane; regional loss became a non-event.
  • Cut mean time to recover from 38 min to 4 min through automated runbooks and brutal game days.
  • Mentor eleven engineers; three now run their own on-call rotations and sleep.
2019 — 2023

Staff SRE, Payments Core · Kestrel Payments

Settlement infrastructure moving the equivalent of a mid-sized country's weekly GDP. Zero tolerance for “eventually consistent”.

  • Led the migration of the ledger to multi-region active-active with no customer-visible downtime.
  • Built the load-shedding layer that absorbed a 6× Black Friday spike while p99 stayed flat.
  • Introduced error budgets; release velocity rose 70% once everyone stopped fearing Fridays.
2016 — 2019

Senior Infrastructure Engineer · Helix Freight Network

Routing, tracking and customs systems for a logistics network spanning 62 countries and several unreasonable legacy mainframes.

  • Rebuilt the event pipeline: 180 TB/day, end-to-end latency under two seconds.
  • Decommissioned a 1998 batch system that nobody dared touch. Ceremony included cake.
2012 — 2016

Network & Systems Engineer · Grayline Datacenters

Where it started: racks, optics, BGP, and learning exactly how many ways a single power strip can ruin a Tuesday.

  • Designed a leaf-spine fabric for 12,000 servers; ran the cutover over one long weekend.
  • Wrote the first version of the capacity model the company still (reluctantly) trusts.

Selected incidents

The Leap-Second Cascade

A single clock step took out three “independent” clusters. Taught a generation of engineers to fear NTP and love monotonic time.

Black Friday, 6× load

Shed the right 4% of traffic at the edge so the other 96% never noticed. Nobody noticed. That was the point.

The Cache That Ate a Region

A thundering herd, a cold cache, and a retry policy written in good faith. Fix: jittered backoff and a circuit breaker with opinions.

Certificate Expiry, 03:12

Not mine, but I wrote the post-mortem. Now every certificate in the company renews itself and files a ticket if it can't.

Toolbox

Linux / eBPFKubernetesGoRustPythonTerraform BGP / AnycastQUIC / HTTP/3PostgreSQLKafkaPrometheusChaos engineeringCapacity planningIncident command

Contact

Engagements are by introduction only. If we share a colleague, ask them. If we don't, a well-written postmortem is the best cover letter.