>cat about.md

# My background and how I work

~/about.md

background

I am a staff engineer. I moved into leadership by taking responsibility for systems that must keep working under peak load. At SCAYLE, I lead checkout and stay close to the backend and platform details that affect reliability.

I do my best work on production problems: noisy alerts, fragile processes, unclear ownership, and systems that fail when traffic arrives. I work on these problems until the system is stable in production.

how it started

My interest in engineering started early. I built WML pages when phones were only beginning to browse the internet. I learned by working inside tight technical limits.

Later I built a text-based game in PHP4 with SMS payments. Mobile operators took about 50% of the revenue. That frustrated me, but it taught me the business side of software. This project is a big part of why I chose software engineering as a career.

current focus

I work on checkout reliability, observability and peak load. I stay close to incidents, find the parts of the system that are hard to understand, and fix the causes of avoidable alert noise.

  • Lead a 7-person checkout team: 4 senior developers, 1 QA and 1 product owner. Run releases, on-call quality, the security backlog and onboarding.
  • Replaced resource-based autoscaling for checkout with traffic-based autoscaling (requests in flight, with KEDA) and capped work per pod. Live in production ahead of Black Friday 2026. In a production load test, it handled 1,000 checkouts per minute for 10 minutes. 99.97% completed, none were dropped, and the service scaled out about 10× on its own.
  • Led checkout through Black Friday 2025, with a peak of about 550 orders per minute and zero checkout downtime during the high-load period.
  • Raised the largest basket from 75 to 500 unique configured items in load tests. Ran 15+ measured load-test rounds with Blackfire and Datadog. Each round found the next bottleneck. One round alone cut a 500-item checkout time by 44%.
  • Designed the plan to separate checkout from the legacy order system, so that its outages and slowness will not stop customers from paying.
  • Designed and built the A/B testing backend and conditional delivery fees. Both are live. Wrote 30+ design documents and technical proposals in total, covering shipping, payments, promotions and API contracts.
  • Owned checkout reliability. Rewrote the monitoring-duty process, checked every SLO against what Datadog really measured, and wrote 15+ incident reports with fixes. Cut response times by 100–270 ms per checkout step.
  • Use AI coding agents daily, with review rules for their output. Wrote an internal guide on where AI gets things wrong.

different kinds of systems

I worked on systems for very different purposes, in several product categories. This changed how I think about tradeoffs and risk.

Real-time betting: Worked on betting systems with a focus on correctness, speed and financial safety.

Warehouse and logistics: Built warehouse workflows and shipping integrations for automotive e-commerce, including DHL, DPD, and UPS.

Startup product work: Moved Atobi from a prototype stack to a Laravel API, a React frontend and mobile apps. The team grew from 3 to 10 engineers.

Search-heavy systems: Rebuilt search for anwalt.de with GraphQL, geolocation, aggregations, and range filters for millions of monthly users.

High-load commerce: Lead checkout at SCAYLE, with a focus on reliability and peak load.

what matters to me

Before SCAYLE, I rebuilt search at anwalt.de and later helped modernize the platform. Before that, I led delivery for Atobi at Solutionlab. The team grew from 3 to 10 engineers.

I care about direct communication and systems that the next engineer can understand. I want software that keeps working and a team that keeps shipping.

Languages
PHPGoTypeScript
Frameworks
LaravelLaravel OctaneSymfonyDoctrineReactRedux
Databases
RedisMySQL
Infrastructure
KubernetesKEDARoadRunnerFrankenPHPKubernetes HPAHelmDockerAWSSQSSNSTerraformDatadogDatadog APMBlackfirek6GitLab CITrivy
Practices
gRPCOpenAPIoasdiffPHPUnitJestCypressPlaywright
AI tools
Claude Code