Platform Engineering
Dr.Job — High-Traffic Recruitment Platform
How I led backend and platform changes that improved availability, latency, delivery safety, and infrastructure efficiency on a live recruitment platform handling approximately two million daily requests.
- Role
- Lead Backend and Platform Engineer
- Period
- Apr 2022 – present
- Status
- Live
Impact at a glance
- Daily requests
- ~2M
- Platform uptime
- 99%+
- Key API latency
- 550ms → 350ms
- PageSpeed score
- 65 → 90+
- Cloud infrastructure cost reduction
- ~37%
- PR review time reduction
- ~75%
Multi-country recruitment platform
Improved from about 80%
Critical API paths
Key user flows
About USD 12.8K → 8.1K/month
Standardised reviews and branching
Context
Dr.Job had grown into a high-traffic, multi-country recruitment platform while much of its core behaviour still lived in a mature Laravel application. The immediate priority was to stabilise production and improve critical user paths. Modernisation had to happen gradually because search, job discovery, and regular feature delivery could not stop.
My responsibility
I owned backend and platform architecture decisions—where work stayed in the Laravel application and where a clearer service boundary was justified. That included service-boundary and API decisions, reliability and performance work on critical paths, production incident follow-through, and the incremental modernisation path.
I also set code review, branching, and release practices, and mentored developers on day-to-day delivery quality. When platform work crossed team lines, I coordinated with product, frontend, DevOps, AI, and vendor teams.
Implementation and outcomes were delivered with those teams. Figures elsewhere describe results I led or shaped, not sole credit for everything shipped on the platform.
Stabilising the operating baseline
I treated monitoring and incident response as part of the platform work, not a separate ops track. When production failed, the recovery path had to be clear, and the same signals had to drive the next fix.
We concentrated on critical API paths that jobseekers and employers hit constantly—especially search and discovery. Caching and data-access changes reduced unnecessary load on those paths. Search performance mattered because filter and discovery latency shaped the product experience directly.
Background workloads were tuned so non-interactive work competed less with user-facing requests. The practical effect was fewer interruptions for users and a more predictable surface for teams still shipping product features on the same platform.