LLM Platform Production Engineer - SRE & Observability

Qube Research & Technologies; Hong Kong; On-site; HKD 600,000 - 900,000; Full time

Job summary

Qube Research & Technologies is seeking an experienced Production Support / SRE professional to own the LLM gateway platform operations in our Hong Kong office. You will diagnose incidents, optimize infrastructure, and collaborate across engineering, data science and business teams to ensure high availability.

You will troubleshoot model-serving backends, provider integrations and observability tooling, including dashboards and alerting, while maintaining documentation and runbooks for rapid

Qualifications

• Experience in production support or SRE within a fast-paced environment.
• Strong ownership of issue resolution end to end.
• Able to communicate clearly with engineers and non-technical users.

Responsibilities

• Provide first- and second-line support for the LLM gateway platform, investigating and resolving issues.
• Monitor and maintain platform infrastructure for availability and performance under growing load.
• Troubleshoot model-serving backends, provider integrations, latency and capacity issues.
• Triage incidents affecting model availability and drive to resolution with vendors.
• Coordinate with platform engineering, cloud infra and end-user teams to resolve incidents.
• Support release management and change processes to keep production stable.

Skills

Linux admin

Windows admin

Python

Bash

PowerShell

SQL

PostgreSQL

SQL Server

Grafana

Prometheus

CI/CD

Terraform

Ansible

Kubernetes

AWS

Job description

Qube Research & Technologies is seeking an experienced Production Support / SRE professional to own the LLM gateway platform operations in our Hong Kong office. You will diagnose incidents, optimize infrastructure, and collaborate across engineering, data science and business teams to ensure high availability.

You will troubleshoot model-serving backends, provider integrations and observability tooling, including dashboards and alerting, while maintaining documentation and runbooks for rapid