Ahrefs is one of the most technically unusual companies in the SEO/web-analytics space, and this SRE posting is a window into an infrastructure philosophy that runs counter to nearly every mainstream trend in the industry.

The stack tells a story. The most striking detail is custom OCaml code at the core of their distributed crawler. OCaml is a niche functional language prized for its type safety, performance, and correctness guarantees — but it has a tiny hiring pool compared to Go, Rust, or Java. Choosing it for a large-scale web crawler signals a team that optimizes for code reliability and raw throughput over ease of hiring. The rest of the stack — Debian, ELK, Puppet — is deliberately boring and battle-tested. No Kubernetes, no Terraform, no cloud-native buzzwords. This is a company that knows exactly what it needs and doesn't chase trends.

Bare-metal servers are the real tell. In 2020, when nearly every startup was going all-in on AWS or GCP, Ahrefs explicitly requires experience with bare-metal infrastructure. Running one of the world's largest web crawlers on owned hardware is a massive cost optimization play. At their scale — crawling billions of pages — cloud compute bills would be astronomical. This posting reveals a company that has done the math and concluded that owning and operating physical servers is a competitive advantage, not a liability.

What this says about their stage and challenges:

• They are mature and profitable enough to invest in physical infrastructure rather than renting cloud capacity.
• The emphasis on daily on-call rotation and 24/7 uptime suggests their crawler is business-critical with zero tolerance for downtime — customers pay for fresh, continuously updated backlink data.
• Offering remote work across Americas time zones for an on-call SRE role shows they're building a follow-the-sun support model, likely because they've hit the limits of a single-timezone ops team.

Green flags: The posting is refreshingly honest. It says "anything else that will solve the task at hand," signaling pragmatism over dogma. Remote-friendly in 2020 (pre-pandemic normalization) was ahead of the curve. The OCaml choice, while unconventional, attracts a self-selecting pool of deeply technical engineers who stay for the intellectual challenge.

Red flags: The OCaml dependency is a double-edged sword. If key engineers leave, institutional knowledge walks out with them. Puppet over more modern configuration management (Ansible, Nix) hints at legacy infrastructure debt. And "daily on-call rotation" without mention of compensation or team size could mean grueling schedules for a small team.