Untitled

Advanced Technical Interview Questions (With Expected Answers)
🐍 Python (Advanced)
1) Explain GIL in Python. When does it become a bottleneck and how do you overcome it?

Expected Answer:

GIL (Global Interpreter Lock) allows only one thread to execute Python bytecode at a time.

Bottleneck in CPU-bound multithreaded programs.

Not a problem for I/O-bound tasks.

Solutions:

Use multiprocessing

Use C extensions (NumPy releases GIL)

Use async programming (for I/O bound)

Use alternative interpreters (Jython, PyPy – limited cases)

2) Difference between asyncio, threading, and multiprocessing?

Expected Answer:

threading: Shared memory, affected by GIL.

multiprocessing: Separate processes, true parallelism, more memory.

asyncio: Single-threaded, cooperative multitasking, event loop.

Best choice depends on workload (CPU-bound vs I/O-bound).

3) What happens internally when you use a Python decorator?

Expected Answer:

Decorator is a higher-order function.

It wraps a function and returns a modified function.

Happens at function definition time.

Uses closures.

functools.wraps preserves metadata.

🚀 FastAPI (Deep)
4) How does FastAPI achieve high performance?

Expected Answer:

Built on Starlette (ASGI framework)

Uses Pydantic for validation

Async support

Runs on Uvicorn (ASGI server)

Automatic OpenAPI generation

5) How do you handle background tasks and dependency injection in FastAPI?

Expected Answer:

BackgroundTasks class

Dependency injection using Depends()

Supports scoped dependencies

Can use middleware for cross-cutting concerns

☁ AWS (Advanced)
6) Difference between EC2 Auto Scaling and Kubernetes HPA?

Expected Answer:

EC2 ASG scales VMs based on CloudWatch metrics.

HPA scales pods based on CPU/custom metrics.

HPA works inside cluster; ASG works at infrastructure level.

Best practice: Use both together.

7) How would you design a highly available system on AWS?

Expected Answer:

Multi-AZ deployment

ALB in front

Auto Scaling

RDS Multi-AZ

S3 for static content

Route53 health checks

Use IAM roles

🐳 Docker (Advanced)
8) Explain difference between CMD and ENTRYPOINT.

Expected Answer:

ENTRYPOINT: fixed executable

CMD: default arguments

ENTRYPOINT + CMD used together

CMD overridden easily

9) What are multi-stage builds?

Expected Answer:

Reduce image size

Separate build and runtime environments

Improves security

☸ Kubernetes (Deep)
10) What happens when a pod crashes?

Expected Answer:

Kubelet detects failure

Restart based on restart policy

ReplicaSet ensures desired replicas

If node fails → rescheduled on another node

11) Explain difference between Deployment, StatefulSet, and DaemonSet.

Expected Answer:

Deployment → stateless apps

StatefulSet → stable identity, persistent storage

DaemonSet → one pod per node

12) How does Kubernetes service discovery work?

Expected Answer:

Kube-DNS / CoreDNS

Service creates DNS entry

ClusterIP virtual IP

kube-proxy manages iptables rules

📦 Terraform (Advanced)
13) What is Terraform state? Why is remote backend important?

Expected Answer:

State file maps infrastructure to config

Tracks resource metadata

Remote backend (S3 + DynamoDB lock)

Collaboration

State locking

Prevent corruption

⚙ Jenkins (Advanced)
14) Difference between scripted and declarative pipelines?

Expected Answer:

Declarative: simpler, structured syntax

Scripted: full Groovy flexibility

Declarative preferred for maintainability

📊 Prometheus
15) How does Prometheus pull metrics?

Expected Answer:

Pull-based model

Scrapes HTTP endpoints

Uses time-series DB

PromQL for queries

Alertmanager for alerts

🐧 Linux (Advanced)
16) Explain what happens when you run a command in Linux.

Expected Answer:

Shell parses command

Fork system call

Exec replaces process image

Parent waits

Uses environment variables

🔐 DevOps Scenario Question (Tough)
17) Your Kubernetes application is randomly restarting. How do you debug?

Expected Answer:

Check kubectl describe pod

Check events

Check logs

Check liveness/readiness probes

Check resource limits (OOMKilled?)

Check node status

Use kubectl top

💻 Tough Coding Questions
🔥 Coding Question 1 (Concurrency + Rate Limiting)
Implement a Thread-Safe Rate Limiter in Python

Design a rate limiter that:

Allows max 5 requests per 10 seconds

Thread-safe

Rejects excess requests

Expected Concepts:

threading.Lock

time window logic

deque

Example usage:

limiter = RateLimiter(5, 10)
if limiter.allow():
    print("Allowed")
    else:
        print("Blocked")

        This tests:

        Concurrency

        Data structures

        Edge case handling

        🔥 Coding Question 2 (System Design + Algorithm)
        Implement an LRU Cache from Scratch (Without Using OrderedDict)

        Requirements:

        O(1) get

        O(1) put

        Fixed capacity

        Use HashMap + Doubly Linked List

        Example:

        cache = LRUCache(2)
        cache.put(1,1)
        cache.put(2,2)
        cache.get(1)  # returns 1
        cache.put(3,3)  # evicts key 2

        Expected Concepts:

        Hash map

        Doubly linked list

        Edge case handling

        Memory efficiency

        🎯 Bonus Extreme Scenario Question
        18) Design a CI/CD pipeline for microservices deployed on Kubernetes using Jenkins and Terraform.

        Expected Answer Should Include:

        Git trigger

        Build Docker image

        Push to ECR

        Terraform infra provisioning

        Helm deployment

        Rollback strategy

        Blue-Green or Canary deployment

        Monitoring with Prometheus

        If you want, I can also give:

        🔥 Mock interview simulation

        🧠 System design whiteboard questions

        💣 Real-world troubleshooting cases

        💻 Full solutions to coding problems

        Tell me which one you want.