11 Years in Tech | Backend | System Design | AI Building secureintent.ai/ DM For Collab & work📩
Based in India
- Tweets 15,467
- Following 838
- Followers 2,766
- Likes 16,338
API throws an error.
No try/catch anywhere.
Express doesn't catch async errors.
Server crashes.
Every user gets a connection reset.
App goes down.
What's missing? 🤔
17
4
43
3,163
Index exists on date column.
Query still scans 50 million rows.
Dashboard takes 40 seconds to load.
Index is there.
Still slow.
Why? 🤔
15
1
30
3,611
Works perfectly.
You add this one line:
const processData = useCallback(() => {}, []);
App breaks.
Data never processed.
Same function name.
Completely different behavior.
Why? 🤔
11
2
26
871
You want all users.
Even ones with zero orders.
But users with no orders
are missing from results.
LEFT JOIN should include everyone.
Why are they disappearing? 🤔
13
1
22
1,245
Email verified.
Token still valid.
Token never expires.
Attacker intercepts token once.
Uses it 1000 times.
Verifies 1000 accounts.
What's missing? 🤔
22
4
45
3,198
The user clicks Submit, then immediately loses internet connectivity.
What happens to the UI, and how would you design this form to handle timeouts, retries, and duplicate submissions correctly?
15
3
29
1,449
User searches for "a".
10 million products scanned.
Server dies.
User searches for "".
Every product returned.
Server dies again.
Same endpoint.
Two different ways to crash it.
What are the fixes? 🤔
9
1
31
1,712
Index on user_id exists.
Table has 50 million rows.
Query takes 30 seconds.
Index is there.
Still slow.
You add another index on created_at.
Still slow.
What's missing? 🤔
14
2
40
5,225
A user double-clicks the Buy button because the first request feels slow.
How can this create duplicate orders, and how would you prevent it from the frontend and backend?
16
33
2,363
User transfers ₹10,000.
First query succeeds.
Internet drops.
Second query never runs.
₹10,000 gone.
Nobody received it.
What's missing? 🤔
24
3
36
1,928
Counter should increment every second.
It doesn't.
Counter is stuck at 1.
Forever.
Cleanup is correct.
Logic looks right.
What's wrong? 🤔
9
1
28
2,975
Subscription created.
End date calculated.
Looks correct.
But every subscription expires on the wrong date.
Can you spot the bug? 🤔
24
1
60
16,225
Works perfectly.
Until this request arrives:
?category=';
DROP TABLE products;
Your entire products table. Gone. Forever.
What's the vulnerability? 🤔
9
4
24
1,093
10 orders = 21 database queries.
100 orders = 201 database queries.
1000 orders = 2001 database queries.
Works in dev.
Kills production.
What's wrong? 🤔
29
6
85
12,209
Frontend only needs name and email.
API returns everything.
Password. Credit card. SSN.
All exposed in the response.
One API call.
Every sensitive field leaked.
What's wrong? 🤔
21
2
54
3,778
Table has 10 million rows.
3 conditions in WHERE clause.
Query takes 45 seconds.
You add an index on department.
Query still takes 40 seconds.
Index exists.
Query is still slow.
Why? 🤔
33
9
169
57,395
Works fine with 100 followers.
User follows 10,000 people.
Query takes 30 seconds.
App freezes.
Same query. Same code.
Just more followers.
What's wrong? 🤔
16
5
47
5,062
Two users register with the same email simultaneously.
Both pass the check.
Both accounts created.
Duplicate email in database.
Authentication broken forever.
No error. No warning.
What's missing? 🤔
34
2
66
10,080
Token is valid.
User is deleted from database.
API returns null.
Frontend crashes.
No error message.
Just blank screen.
What are the 2 things missing? 🤔
19
2
54
2,568
1 item in stock.
1000 users order simultaneously.
Stock check passes for all 1000.
1000 orders created.
Stock goes to -999.
How do you fix this? 🤔
53
8
133
16,771
Dashboard loads fine at 9 AM.
100 people open it at 9:01 AM.
Same query runs 100 times.
Database dies.
Query takes 8 seconds.
100 users × 8 seconds = 800 seconds of DB work.
What's the fix? 🤔
34
7
64
5,981
Frontend Interview:
React.memo should prevent re-renders.
But Button re-renders every time.
Why is memo not working? 🤔
27
3
122
21,990
Backend Interview:
10,000 users.
Each email takes 100ms.
Total time: 1000 seconds.
16 minutes to send all emails.
One change makes it 10x faster.
Same result. Same emails.
What's the fix? 🤔
69
14
329
128,958
Attacker sends: filename: "../../../config/database.js"
Your database config just got overwritten.
Credentials exposed.
Server compromised.
No hacking tools needed.
Just a filename.
What's the vulnerability? 🤔
21
4
113
11,248
Backend Interview:
Attacker changes the header:
x-user-id: 1 -> sees User 1's balance
x-user-id: 2 -> sees User 2's balance
x-user-id: 3 -> sees User 3's balance
Every account. Every balance.
No password needed.
Just change a number.
What's the fix? 🤔
29
3
56
6,765
User clicks button.
Navigates away before API responds.
Component unmounts.
API responds.
State updates on unmounted component.
React throws a warning.
Memory leak.
What's the fix? 🤔
9
2
26
1,006
Works perfectly on page 1.
Page 1000 takes 30 seconds.
Page 10000 crashes the server.
Pagination is supposed to make it faster.
It's making it slower.
Why? 🤔
22
3
55
4,216
Traffic suddenly increases to 10,000 requests/minute. Why can this code exhaust the connection pool?
How would you fix it?
23
1
26
595
Frontend Interview:
Every keystroke.
expensiveSearch runs.
UI freezes.
User types "javascript".
10 characters.
10 expensive calculations.
One hook fixes this.
What is it? 🤔
22
3
43
934
Backend Interview:
User clicks "Place Order."
Waits 8 seconds.
Email sending -> 3 seconds
Inventory update -> 2 seconds
Warehouse notification -> 3 seconds
User doesn't care about any of this.
They just want their order confirmed.
How do you make this instant? 🤔
14
1
30
708
Every button click.
ExpensiveComponent re-renders.
Even though nothing changed for it.
App gets slower with every click.
One line fix.
What is it? 🤔
29
2
52
3,611
DB insert succeeds.
Cache update fails.
Search never indexed.
Data exists in database.
Cache is stale.
Search returns nothing.
No error shown to user.
Just inconsistent data everywhere.
What's the fix? 🤔
28
1
47
1,769
Expected: "John"
Actual: undefined
Same function.
Same code.
Different result.
Why did this disappear? 🤔
19
1
46
2,684
Backend Interview :
Payment gateway sends a webhook.
Your server is slow.
Gateway retries 3 times.
Same payment processed 3 times.
Customer charged 3 times.
No error. No duplicate warning.
What's missing? 🤔
15
2
47
2,196
Backend Interview:
No authentication.
No authorization.
Just pass any userId.
But the real problem is something else entirely.
Can you spot it? 🤔
21
3
73
16,400
Backend Interview:
Attacker creates their own JWT.
Sets role: "admin".
Gets access to every user in your database.
Can you spot why? 🤔
40
8
223
53,756
Frontend Interview:
3 bugs in this code.
Looks completely fine.
Every React developer has written this.
Can you find all 3? 🤔
47
9
188
58,914
Backend Interview:
User searches "a".
Database scans 10 million products.
Server dies.
Every. Single. Time.
What's wrong and how do you fix it? 🤔
83
29
458
103,474
Backend Roadmap (Part 11/11): AI Engineering
AI won't replace backend engineers.
Backend engineers who use AI will.
This is the final skill - and possibly the most important one for 2026 and beyond.
In this post, I covered everything you need to build intelligent backend systems:
# LLM APIs - Use managed models via API. OpenAI, Anthropic, Gemini, Mistral, Groq. Choose based on cost, latency, quality & context window. Use system prompts, control temperature, stream for better UX.
# MCP (Model Context Protocol) - The standard way to connect LLMs with your tools and data securely. Think of it as USB-C for AI apps. Standardized, reusable, observable tool calling.
# AI Agents - LLMs that don't just respond - they plan, reason, and act. The Agent Loop: Observe -> Think -> Act -> Remember. Frameworks: LangChain, LlamaIndex, AutoGen, CrewAI, Haystack.
# Vector Databases & Embeddings - Store text as vectors for semantic search. Chunk -> Embed -> Store -> Retrieve. Powers RAG, recommendations, deduplication. Tools: Pinecone, Weaviate, Qdrant, Chroma.
# RAG (Retrieval Augmented Generation) - Combine LLMs with your own data. Retrieve relevant context -> Augment the prompt -> Generate accurate answers. Prevents hallucinations, adds citations, keeps responses grounded.
Evaluation Metrics that matter:
- Accuracy -> is the answer correct?
- Faithfulness -> based only on provided context?
- Relevance -> how relevant is the response?
- Coherence -> well-structured & easy to read?
- Latency + Cost -> can it scale affordably?
Real Challenges to plan for:
- Hallucinations -> use RAG + guardrails
- Prompt injection -> sanitize inputs
- High latency/cost -> cache + optimize
- Model drift -> version & evaluate continuously
- Observability gaps -> log inputs, outputs & errors
Data is your moat. Quality + Context + Execution = Differentiation.
Start small, ship useful AI features, learn, measure, and iterate fast.
That's a wrap on all 11 Backend Skills for 2026!
1) API Design
2) Authentication & Authorization
3) Databases
4) Caching
5) Event-Driven Systems
6) Concurrency & Async Programming
7) Distributed Systems
8) Security
9) Observability
10) Cloud & Deployment
11) AI Engineering
Save this series. Share it with someone learning backend. 🔖
Which skill are you going to deep-dive into first? 👇
Backend Roadmap (Part 10/11): Cloud & Deployment
Great code means nothing if it never ships reliably.
How you deploy is just as important as what you build.
Ship fast. Scale smart. Recover faster.
In this post, I covered everything modern deployment demands:
# Docker - Containerize your app with all its dependencies. Build once, run anywhere. Consistent environments from dev to prod. Lightweight, portable, version-controlled.
# Kubernetes - Orchestrate containers at scale. Auto-scaling, self-healing, rolling updates, service discovery, load balancing. The standard for production-grade deployments.
# Serverless - Run code without managing servers. Pay only for what you use. Perfect for APIs, cron jobs, file processing, and spiky unpredictable workloads. (AWS Lambda, Cloud Run)
# GitOps - Git is your single source of truth. Declarative deployments, automatic sync to target environment. ArgoCD + FluxCD make this seamless.
CI/CD Pipeline - Automate everything: build -> test -> scan -> deploy. GitHub Actions, GitLab CI, Jenkins, CircleCI. Deliver frequently and reliably.
Deployment Strategies - pick the right one:
- Rolling Update -> gradual, zero downtime
- Blue/Green -> instant switch, easy rollback
- Canary ->test with small % of users first
- Recreate -> stop all, deploy new (with downtime)
Cloud Providers at a glance:
- AWS -> EC2, EKS, Lambda, S3, RDS, CloudWatch
- GCP -> GKE, Cloud Run, BigQuery, Cloud Monitoring
- Azure -> AKS, Functions, Blob Storage, Azure Monitor
Automate everything. Keep it repeatable. Observe everything. Fail fast, recover faster. Ship small, ship often.
Pro Tip: Focus on building reliable systems - not just deploying code.
This is Part 10 of my "11 Skills Every Backend Developer Should Master in 2026" series.
One more to go -> AI Engineering (LLMs, MCP, AI Agents, Vector DBs, RAG)
Which deployment strategy does your team use? And what's your cloud of choice? 👇
38
11
114
4,882
Backend Interview:
50 million orders in database.
Someone calls this endpoint.
Server runs out of memory.
App crashes.
No error handling.
No pagination.
No limit.
One API call brought down production.
What's missing? 🤔
24
4
59
30,046
Backend Roadmap (Part 10/11): Cloud & Deployment
Great code means nothing if it never ships reliably.
How you deploy is just as important as what you build.
Ship fast. Scale smart. Recover faster.
In this post, I covered everything modern deployment demands:
# Docker - Containerize your app with all its dependencies. Build once, run anywhere. Consistent environments from dev to prod. Lightweight, portable, version-controlled.
# Kubernetes - Orchestrate containers at scale. Auto-scaling, self-healing, rolling updates, service discovery, load balancing. The standard for production-grade deployments.
# Serverless - Run code without managing servers. Pay only for what you use. Perfect for APIs, cron jobs, file processing, and spiky unpredictable workloads. (AWS Lambda, Cloud Run)
# GitOps - Git is your single source of truth. Declarative deployments, automatic sync to target environment. ArgoCD + FluxCD make this seamless.
CI/CD Pipeline - Automate everything: build -> test -> scan -> deploy. GitHub Actions, GitLab CI, Jenkins, CircleCI. Deliver frequently and reliably.
Deployment Strategies - pick the right one:
- Rolling Update -> gradual, zero downtime
- Blue/Green -> instant switch, easy rollback
- Canary ->test with small % of users first
- Recreate -> stop all, deploy new (with downtime)
Cloud Providers at a glance:
- AWS -> EC2, EKS, Lambda, S3, RDS, CloudWatch
- GCP -> GKE, Cloud Run, BigQuery, Cloud Monitoring
- Azure -> AKS, Functions, Blob Storage, Azure Monitor
Automate everything. Keep it repeatable. Observe everything. Fail fast, recover faster. Ship small, ship often.
Pro Tip: Focus on building reliable systems - not just deploying code.
This is Part 10 of my "11 Skills Every Backend Developer Should Master in 2026" series.
One more to go -> AI Engineering (LLMs, MCP, AI Agents, Vector DBs, RAG)
Which deployment strategy does your team use? And what's your cloud of choice? 👇
Backend Roadmap (Part 9/11): Observability 👁️
Your system is down.
Users are complaining.
You're staring at the screen -blind.
You can't fix what you can't see.
Observability is what separates engineers who guess from engineers who know.
In this post, I broke down all 4 pillars:
Logs - What happened? Discrete events with context. Always use structured JSON logs with trace_id, user_id, and service name. Tools: ELK Stack, Loki, Splunk, Datadog.
Metrics - What is happening? Numeric data over time.
Track the 4 Golden Signals (Google SRE):
- Latency -> how long requests take
- Traffic -> how many requests
- Errors -> rate of failures
- Saturation -> how full your system is
Tools: Prometheus, Grafana, CloudWatch, Datadog.
Tracing - Why did it happen? Follow a single request across every service with a Trace ID. Identify exactly where latency lives. Tools: Jaeger, Zipkin, AWS X-Ray, Datadog APM.
OpenTelemetry (OTel) - One open standard to collect all three. Vendor-neutral, future-proof. One SDK ->logs + metrics + traces -> any backend.
Observability Maturity - where are you?
- Level 0 -> No logs (flying blind)
- Level 1 -> Logs only (hard to correlate)
- Level 2 -> Logs + Metrics (better visibility)
- Level 3 -> Logs + Metrics + Traces (deep insights)
- Level 4 -> Full Observability (proactive, data-driven)
* Observe -> Understand -> Improve.
Instrument everything that matters. Correlate logs, metrics & traces. Build systems you can see, trust & scale.
This is Part 9 of my "11 Skills Every Backend Developer Should Master in 2026" series.
Next up -> Cloud & Deployment (Docker, Kubernetes, Serverless, GitOps) ☁️
What's your current observability stack? Are you using OpenTelemetry yet? 👇
37
5
79
8,567