Role Overview: We are seeking an AI Quality & Reliability Engineer to own the "Production Readiness" of our cloud-based AI solutions. This hybrid role combines automated software testing, AI model evaluation, and Site Reliability Engineering (SRE). You will build the automated frameworks that validate our AI outputs and ensure the underlying Azure/AWS infrastructure is resilient, performant, and compliant with banking standards.
Key Responsibilities:
Key Responsibilities:
- AI Quality Automation: Design and execute automated testing frameworks for AI services (e.g., Azure OpenAI, AWS Bedrock). This includes testing for model hallucinations, accuracy, and "AI Content Safety" latency.
- Resiliency Engineering (SRE): Implement "Chaos Engineering" and load testing to ensure web/mobile backends can handle banking-scale traffic. Maintain high availability through automated recovery scripts.
- Automated Regression: Build CI/CD-integrated test suites using Python that validate both the application logic and the infrastructure state (IaC validation).
- Observability & SLIs: Define and monitor Service Level Indicators (SLIs) and Objectives (SLOs). Set up advanced alerting in Azure Monitor or AWS CloudWatch to catch performance degradation before users do.
- Security & Compliance Testing: Automate security scans and compliance checks to ensure all AI data handling meets strict banking data residency and privacy protocols.
- Automation Stack: High proficiency in Python (for AI testing) and framework automation (PyTest, Selenium, or Robot Framework).
- Cloud Infrastructure: Strong hands-on experience with Azure or AWS, specifically regarding networking, scaling, and serverless reliability.
- AI/ML Understanding: Understanding of Prompt Engineering and how to evaluate AI model outputs (RAG evaluation, ROUGE/BLEU scores, or custom LLM-benchmarks).
- Monitoring Tools: Experience with Grafana, Prometheus, or native cloud monitoring tools to build real-time reliability dashboards.
- FinOps Awareness: Ability to identify "expensive" failing tests or inefficient cloud resource usage during the testing phase.
- Languages: Python (Mandatory), Bash scripting.
- Tools: GitHub Actions (CI/CD), Terraform (reading/validating), K6 or JMeter (Performance).
- AI Frameworks: DeepEval, Ragas, or LangSmith (for automated AI evaluation).
Employer-sourced description · source: halian.com · captured 3h ago
