PUA
People Usually Ask
AWS & Cloud Architecture
Q: Explain the Serverless Triad and how it scales.
Answer: It consists of API Gateway (Entry/Auth), Lambda (Logic), and DynamoDB (Data). Scaling is handled by AWS: Gateway throttles requests, Lambda spins up new containers per request, and DynamoDB uses on-demand capacity. To optimize, I use Provisioned Concurrency for Lambda and Single-Table Design for DynamoDB to minimize join-like latencies.
Q: Describe a complex system you've built on AWS. What services did you use and why?
Answer: I built a real-time analytics platform using Kinesis Data Streams for ingestion, Lambda for processing, S3 for storage, and QuickSight for visualization. I chose Kinesis over SQS for real-time requirements, Lambda for auto-scaling, and partitioned S3 by date for cost-effective querying with Athena.
Q: How do you ensure high availability and fault tolerance for an AWS application?
Answer: I deploy across multiple AZs using Auto Scaling Groups, implement Circuit Breakers with exponential backoff, use RDS Multi-AZ for database failover, and CloudFront for global distribution. I also implement health checks and use Route 53 for DNS failover between regions.
Q: Explain security considerations for cross-account S3 access.
Answer: Use IAM Cross-Account Roles with least privilege, implement S3 Bucket Policies with condition keys (aws:SourceAccount), enable S3 Block Public Access, use VPC Endpoints for private access, and implement CloudTrail for audit logging. Never use long-term access keys.
Q: How would you set up a CI/CD pipeline for an AWS-hosted React application?
Answer: Use CodeCommit/GitHub for source, CodeBuild for building React app, S3 for hosting static files, CloudFront for CDN, and CodePipeline to orchestrate. Include automated testing, security scanning, and blue/green deployments with CloudFormation for infrastructure as code.
Advanced React & Performance
Q: How do you optimize a sluggish React application? What are common useEffect pitfalls?
Answer: I use React DevTools Profiler to identify bottlenecks, implement React.memo for component memoization, use useMemo/useCallback for expensive calculations, and avoid missing dependencies in useEffect. Common pitfalls: infinite re-renders from missing deps, not cleaning up subscriptions, and running effects on every render.
Q: Explain React.memo, useMemo, and useCallback use cases to prevent re-renders.
Answer: React.memo prevents component re-renders when props haven't changed. useMemo caches expensive calculations between renders. useCallback memoizes function references to prevent child re-renders. Use sparingly - only when profiling shows actual performance issues, not preemptively.
Q: How do you optimize rendering for large lists (virtualization)?
Answer: I use react-window or react-virtualized to render only visible items. For dynamic heights, I use VariableSizeList. I also implement intersection observer for infinite scrolling and use React.memo on list items to prevent unnecessary re-renders of individual rows.
Q: Explain React Fiber and the reconciliation algorithm.
Answer: React Fiber is the reconciliation engine that breaks work into chunks, enabling time-slicing and prioritization. It uses a virtual stack to pause/resume work, prioritizing user interactions over background updates. The diffing algorithm compares virtual DOM trees using keys and component types for efficient updates.
Q: When to use Context API vs Redux/Zustand?
Answer: Use Context API for theme, auth, or infrequent updates. Use Redux/Zustand for complex state logic, time-travel debugging, or frequent updates across many components. Context causes all consumers to re-render, while Redux uses selectors for granular subscriptions.
Q: Explain Server-Side Rendering (SSR) vs Client-Side Rendering (CSR) benefits.
Answer: SSR provides better SEO, faster initial page load, and works without JavaScript. CSR offers better user interactions after load and reduced server load. Modern solutions like Next.js use hybrid approaches with static generation and incremental static regeneration for optimal performance.
Backend & System Design
Q: How would you design a dashboard that scales to thousands of concurrent users?
Answer: I'd use CDN for static assets, Redis for session storage, WebSocket clusters with sticky sessions, database read replicas, implement pagination/virtualization on frontend, use GraphQL with DataLoader for efficient queries, and horizontal pod autoscaling in Kubernetes.
Q: How do you handle API security including rate limiting and authentication?
Answer: I implement JWT with refresh tokens, use API Gateway for rate limiting (sliding window), validate all inputs with Joi/Zod, implement CORS properly, use helmet.js for security headers, and add request signing for sensitive operations. Rate limiting uses Redis with user-based keys.
Q: Explain blue/green vs rolling deployment trade-offs.
Answer: Blue/Green: Instant rollback, zero downtime, but requires 2x infrastructure. Rolling: Resource efficient, gradual deployment, but slower rollback and potential version mixing. I use blue/green for critical services and rolling for cost-sensitive applications with good monitoring.
Q: How do you stop bots from scraping a public API?
Answer: Implement rate limiting with progressive delays, use CAPTCHA for suspicious patterns, require API keys with usage quotas, analyze User-Agent patterns, implement IP reputation scoring, use CloudFlare Bot Management, and add honeypot endpoints to detect scrapers.
Q: How do you ensure data consistency between SQL and NoSQL databases?
Answer: I use the Saga pattern for distributed transactions, implement event sourcing with eventual consistency, use CDC (Change Data Capture) to sync data, implement compensating transactions for rollbacks, and use idempotent operations with unique request IDs.
Q: What are RESTful API best practices?
Answer: Use proper HTTP methods (GET, POST, PUT, DELETE), implement consistent URL patterns (/users/{id}), return appropriate status codes, use HATEOAS for discoverability, implement versioning (v1, v2), add pagination for collections, and include proper error responses with meaningful messages.
Behavioral & Leadership
Q: What is your philosophy on reusable components vs simplicity and premature optimization?
Answer: I follow the "Rule of Three" - abstract after the third use case. I prioritize composition over inheritance, keep components focused on single responsibilities, and avoid premature abstraction. I measure reusability success by reduced maintenance overhead, not just code reuse metrics.
Q: How do you handle technical debt in a growing codebase?
Answer: I categorize debt by impact vs effort, allocate 20% sprint capacity for debt reduction, implement automated code quality gates, use ADRs (Architecture Decision Records) to document trade-offs, and refactor incrementally using the Strangler Fig pattern for large changes.
Q: Describe a major technical challenge you faced. How did you diagnose and solve it?
Answer: I faced a memory leak in a Node.js service causing crashes. I used clinic.js and heapdump to identify retained objects, found unclosed database connections in error paths. I implemented connection pooling, added proper error handling, and created monitoring dashboards to prevent recurrence.
Q: How do you perform effective code reviews and mentor junior engineers?
Answer: I focus on architecture and logic over style (automated by linters), ask "why" questions to encourage thinking, provide specific examples and resources, use pair programming for complex features, and create learning paths with gradually increasing responsibility.
Q: How do you structure a large React codebase for scalability?
Answer: I use feature-based folder structure over type-based, implement barrel exports for clean imports, use absolute imports with path mapping, create shared component libraries, implement lazy loading for routes, and maintain strict TypeScript with proper interfaces.