Technical practice
Designing Multi-Application Azure Environments
What to coverExplain your approach to managing a portfolio of distinct AI applications within a single Azure subscription or across multiple subscriptions. Focus on how you maintain environment parity between staging and production while ensuring security and resource isolation.
Implementing IaC for Complex AI Workloads
What to coverDescribe your workflow for using Terraform and Ansible to manage infrastructure. Discuss how you handle state management, modularity, and GitOps integration to ensure that infrastructure changes are reproducible and auditable.
Managing Stateful Workloads in Kubernetes
What to coverDiscuss the challenges of running stateful AI services, such as vector databases or RAG pipeline backends, on AKS. Explain how you handle persistent storage, data consistency, and graceful pod disruption during cluster scaling or upgrades.
Designing Disaster Recovery for Heterogeneous Data Stores
What to coverDetail your strategy for architecting DR plans across different database types like PostgreSQL, MongoDB Atlas, and CosmosDB. Address how you define and validate RTO/RPO targets in a cloud-native environment.
Building Observability for AI Pipelines
What to coverExplain how you would set up an end-to-end observability stack using Prometheus, Grafana, and ELK/OpenSearch for AI services. Focus on how you monitor inference latency, model performance, and infrastructure health simultaneously.
Cloud Cost Governance and Optimization
What to coverProvide examples of how you have performed cloud cost right-sizing and reserved capacity planning in a production environment. Explain the metrics you use to justify infrastructure spend versus performance requirements.
Securing AI Infrastructure
What to coverDescribe your approach to implementing Entra ID access control, secrets management, and encryption for sensitive data in AI applications. How do you ensure compliance with PI data handling requirements while maintaining operational agility?
Automating Operational Workflows with Python and Bash
What to coverProvide an example of a complex operational task you automated using Python or Bash to move from reactive support to self-healing infrastructure. Focus on the design principles that make your scripts production-ready and maintainable.
Scaling Vector Database Infrastructure
What to coverDiscuss the infrastructure considerations for deploying vector databases like pgvector or Qdrant at scale. Address how you handle indexing, memory requirements, and query performance for embedding pipelines.
Behavioural practice
Managing Conflicting Priorities Across Multiple Projects
What to coverDescribe a situation where you had to manage infrastructure for multiple live applications simultaneously. How did you prioritize tasks, handle resource contention, and ensure that a failure in one application did not impact the others?
Proactive Ownership of Infrastructure
What to coverGive an example of when you identified a systemic infrastructure issue before it became a critical incident. Explain the steps you took to own the resolution and prevent future occurrences without waiting for a ticket.
Collaborating with Non-Infrastructure Teams
What to coverDescribe a time you worked closely with AI architects or research teams to deploy a new model or service. How did you bridge the gap between their requirements and your infrastructure constraints to ensure a successful, scalable deployment?
Handling Production Incidents in High-Stakes Environments
What to coverShare an experience where you had to lead the response to a production outage in a critical system. How did you manage the incident, communicate with stakeholders, and perform a post-mortem to improve platform resilience?