@nanimma
Account based inIndia
About this account
- Account based in
- India
- Connected via
- India Android App
Account-level information from X, not a live location or the device used for a specific post.
Yet to find Myself
అవసరమా!
Joined July 2009
- Tweets3.2K
- Following186
- Followers45
- Likes1K
Narendra N retweeted
Learn AI for free directly from top companies
1 - Anthropic:
anthropic.skilljar.com
2 - Google:
grow.google/ai
3 - Meta:
ai.meta.com/resources/
4 - NVIDIA:
developer.nvidia.com/cuda
5 - Microsoft:
learn.microsoft.com/en-us/tr…
6 - OpenAI:
academy.openai.com
7 - IBM:
skillsbuild.org
8 - AWS:
skillbuilder.aws
9 - DeepLearning AI:
deeplearning.ai
10 - Hugging Face:
huggingface.co/learn
Comment "AI" for more resources.
Like, Retweet & Bookmark for future updates.
Narendra N retweeted
Claude just dropped 13 FREE AI courses (with certificates).
No $500 course needed.
No “guru” required.
Just real skills — straight from Anthropic.
Here’s the full list:
👇
1. Claude 101
lnkd.in/gCPUQsRg
2. AI Fluency: Frameworks & Foundations
lnkd.in/gS6ceZ_M
3. Introduction to Agent Skills
lnkd.in/g_wWNiEb
4. Building with the Claude API
lnkd.in/gDr5K_B4
5. Claude Code in Action
lnkd.in/g9wWZbK9
6. Introduction to Model Context Protocol
lnkd.in/gAj5HqMY
7. MCP: Advanced Topics
lnkd.in/g3eDwBFY
8. AI Fluency for Students
lnkd.in/gKKujHGG
9. AI Fluency for Educators
lnkd.in/gVcKnuhA
10. Teaching AI Fluency
lnkd.in/g9P4gJFM
11. AI Fluency for Nonprofits
lnkd.in/gpsm_BVf
12. Claude with Amazon Bedrock
lnkd.in/gbfPjSFt
13. Claude with Google Vertex AI
lnkd.in/gvVgB4Ub
—
If you go through even HALF of these…
You’ll be ahead of 95% of people using AI.
Most people won’t.
Because they’re still:
• Watching random YouTube videos
• Buying overpriced courses
• “Learning AI” without actually building
Don’t be that person.
Do this instead:
1. Save this post (you’ll come back to it)
2. Pick 1 course → start today
3. Share it with someone who needs this
Free. Practical. No excuses.
Narendra N retweeted
How to Learn Claude in 12 Days
(This will save you 2+ hours every day)
Most people use Claude like a chatbot.
Ask a question. Get an answer. Repeat.
Big mistake.
Claude is becoming an operating system for knowledge work.
Here's a 12-day roadmap based on real Claude workflows:
☀️ DAY 1: Build CLI Tools
Goal: Turn ideas into working tools.
→ Open Claude Code
→ Describe a tool in plain English
→ Let Claude generate and run it
No frameworks. No boilerplate. Just describe → build.
☀️ DAY 2: Create an MCP Server
Goal: Give Claude new capabilities.
→ Pick a tool or API you use often
→ Define endpoints and authentication
→ Let Claude generate the server
Now Claude can interact with your systems directly.
☀️ DAY 3: Build a Personal RAG
Goal: Make Claude remember your knowledge.
→ Upload notes, docs, and PDFs
→ Connect Google Drive, Notion, or Gmail
→ Ask questions across everything
Your second brain becomes searchable.
☀️ DAY 4: Understand Any Codebase
Goal: Learn unfamiliar code fast.
→ Drop in a GitHub repo
→ Ask for architecture breakdowns
→ Follow guided walkthroughs
Hours of reading become minutes.
☀️ DAY 5: Practice Interviews
Goal: Improve faster than studying alone.
→ Pick a role
→ Choose an interview type
→ Let Claude act as interviewer
Get feedback after every answer.
☀️ DAY 6: Automate the Browser
Goal: Stop doing repetitive web tasks.
→ Tell Claude what you need
→ Let it navigate websites
→ Complete tasks automatically
Research becomes delegation.
☀️ DAY 7: Tailor Every Resume
Goal: Apply smarter.
→ Paste a job description
→ Upload your resume
→ Ask Claude to optimize for ATS
Customized applications in minutes.
☀️ DAY 8: Schedule Cloud Routines
Goal: Work while you sleep.
→ Create recurring tasks
→ Schedule reports or research
→ Wake up to completed work
Productivity without active effort.
☀️ DAY 9: Design Faster
Goal: Go from idea to visual.
→ Describe a landing page
→ Describe a pitch deck
→ Let Claude generate drafts
Blank-page syndrome disappears.
☀️ DAY 10: Run Multi-Agent Reviews
Goal: Get multiple perspectives.
→ Assign different review roles
→ Let agents inspect code independently
→ Compare recommendations
One review becomes many.
☀️ DAY 11: Use Auto Mode
Goal: Delegate entire workflows.
→ Give Claude a specification
→ Let it plan, code, test, and review
→ Ship end-to-end
You manage outcomes, not steps.
☀️ DAY 12: Build Custom Skills
Goal: Create reusable expertise.
→ Save recurring workflows
→ Turn prompts into commands
→ Reuse them forever
Your best workflows become one-click actions.
Most people use Claude for answers.
Power users use Claude for systems.
The difference isn't intelligence.
It's setup.
Do this instead of scrolling:
→ Save this post
→ Start Day 1 today (10 minutes)
→ Complete one day at a time
→ Come back in 12 days
You'll never use AI the same way again. 🚀
Narendra N retweeted
Learn AI for free directly from top companies
1 - Anthropic:
anthropic.skilljar.com
2 - Google:
grow.google/ai
3 - Meta:
ai.meta.com/resources/
4 - NVIDIA:
developer.nvidia.com/cuda
5 - Microsoft:
learn.microsoft.com/en-us/tr…
6 - OpenAI:
academy.openai.com
7 - IBM:
skillsbuild.org
8 - AWS:
skillbuilder.aws
9 - DeepLearning AI:
deeplearning.ai
10 - Hugging Face:
huggingface.co/learn
Comment "AI" for more resources.
Like, Retweet & Bookmark for future updates.
Narendra N retweeted
If you're serious about becoming a Senior Backend Engineer in 2026, master this progression:
Level 1: Postman / Bruno
Test APIs properly before your users become your QA team. Contracts matter from day one.
Level 2: Redis
Speed is easy. Cache invalidation is the real challenge. Learn to weigh trade-offs over cleverness.
Level 3: PostgreSQL
Data modeling, indexing, transactions the foundation of real backend thinking. Schema design is system design.
Level 4: Kafka
Async systems are powerful, but retries, ordering, and idempotency separate juniors from seniors. Failure is a feature you design for.
Level 5: Docker
"Works on my machine" should have ended in your junior years. Consistency beats convenience every time.
Level 6: OpenTelemetry
Trace requests across services and distributed systems finally make sense end-to-end. Observability isn't optional it's essential.
Level 7: Grafana
Dashboards that show latency, errors, and throughput not just pretty graphs. If you can't measure it, you can't improve it.
Level 8: Prometheus
Metrics that force you to think in SLOs and system health, not just features. Reliability outpaces feature velocity in the long run.
Level 9: k6
Load test your "scalable" backend. Watch it break. Then fix it. Scale is a verb, not an adjective.
Level 10: Terraform
Senior engineers don't just write code. They own the infrastructure it runs on. Infrastructure is code. Ownership is culture.
The pattern:
• Foundation: Postman/Bruno, PostgreSQL, Redis
• Resilience: Kafka, Docker, k6
• Observability: OpenTelemetry, Grafana, Prometheus
• Ownership: Terraform
Master one layer before rushing the next.
(save it)
Narendra N retweeted
If I had to start from zero today:
This is the exact roadmap I'd follow.
☑ Phase 1: Learn how software is built.
Before Claude.
Before AI.
Before agents.
Learn:
→ Python
→ APIs
→ Git
→ GitHub
→ JSON
→ Terminal
Most people skip this.
Then wonder why Claude's code breaks.
━━━━━━━━━━━━━━━
☑ Phase 2: Learn Claude properly.
Not prompting.
Claude.
Learn:
→ Projects
→ Artifacts
→ Memory
→ Knowledge Files
→ Claude Desktop
Most people use 5% of Claude.
That's the problem.
━━━━━━━━━━━━━━━
☑ Phase 3: Learn Claude Code.
Start simple.
Open a project.
Then ask:
→ Explain this codebase
→ Find bugs
→ Refactor this file
→ Create tests
→ Generate documentation
Don't generate apps.
Understand workflows.
━━━━━━━━━━━━━━━
☑ Phase 4: Learn Claude Code Structure.
This changes everything.
Create:
/docs
/prd
/tasks
/rules
/context
Every serious AI engineer eventually learns:
Bad context = bad output.
Good context = magic.
Claude becomes powerful when it understands:
→ Product requirements
→ Coding rules
→ Project structure
→ Existing decisions
━━━━━━━━━━━━━━━
☑ Phase 5: Learn Context Engineering.
The new skill nobody talks about.
Learn:
→ CLAUDE.md
→ Project Instructions
→ Rule Files
→ Context Compression
→ Memory Systems
Prompting is temporary.
Context is infrastructure.
━━━━━━━━━━━━━━━
☑ Phase 6: Build AI Systems.
Projects:
→ AI SaaS
→ AI CRM
→ AI Research Agent
→ AI Content Engine
→ AI Customer Support Agent
Don't build tutorial projects forever.
Build products.
━━━━━━━━━━━━━━━
☑ Phase 7: Advanced Claude.
Learn:
→ MCP
→ Agent Workflows
→ Multi-Agent Systems
→ Tool Calling
→ RAG
→ Evaluations
This is where senior AI engineers operate.
━━━━━━━━━━━━━━━
Resources I'd use:
→ Anthropic Docs
→ Claude Documentation
→ GitHub Repositories
→ DeepLearning.AI
→ OpenAI Docs
And honestly:
Stop buying expensive AI courses.
Follow @vishisinghal_ for more
🔁 Repost it to share in your network!!
Narendra N retweeted
If you're serious about starting with system design, learn these 19 concepts (save this now):
1 System Design Concepts
↳ newsletter.systemdesign.one/…
2 How to Crack the System Design Interview
↳ newsletter.systemdesign.one/…
3 Computer Science Stack Simply Explained
↳ newsletter.systemdesign.one/…
4 High Availability - A Deep Dive
↳ newsletter.systemdesign.one/…
5 Modular Monolith Architecture
↳ newsletter.systemdesign.one/…
6 How RPC Works
↳ newsletter.systemdesign.one/…
7 How JWT Works
↳ newsletter.systemdesign.one/…
8 How Does HTTPS Work
↳ newsletter.systemdesign.one/…
9 How Bloom Filters Work
↳ systemdesign.one/bloom-filte…
10 How Consistent Hashing Works
↳ systemdesign.one/consistent-…
11 How Service Discovery Works
↳ systemdesign.one/what-is-ser…
12 API Versioning - A Deep Dive
↳ newsletter.systemdesign.one/…
13 Deployment Patterns
↳ newsletter.systemdesign.one/…
14 How Idempotent API Works
↳ newsletter.systemdesign.one/…
15 Saga Design Pattern
↳ newsletter.systemdesign.one/…
16 How Databases Keep Passwords Securely
↳ newsletter.systemdesign.one/…
17 How DNS Works
↳ newsletter.systemdesign.one/…
18 How Websockets Work
↳ newsletter.systemdesign.one/…
19 Distributed Systems 101
↳ newsletter.systemdesign.one/…
(What else should make this list?)
——
👋 PS - Want my System Design Playbook for FREE?
Click the link below to join my newsletter right now:
→ newsletter.systemdesign.one/…
(200K+ software engineers have already signed up.)
———
💾 Save this for later & RT to help other software engineers ace system design.
👤 Follow @systemdesingone + turn on notifications.
Narendra N retweeted
The biggest lie in AI right now is “AI will replace junior developers”
Here’s what’s actually happening:
1. The “Autocomplete” Trap
AI generates 500 lines in 30 seconds. But it can’t tell you if those 500 lines match your product constraints, edge cases, or failure modes.
It ships something. Juniors must learn to ship the right thing.
2. The “Integration” Wall
Real work lives in glue code:
- auth + permissions
- data models + migrations
- caching + retries
- deployments + envs
- observability + on-call realities
Juniors become valuable when they can stitch the system together without breaking it.
3. The “Hallucination Tax”
Every AI-generated feature comes with hidden costs:
- subtle bugs
- missing validations
- security footguns
- flaky behavior under load
The person who can read code, debug fast, and write tests is a good developer today.
4. The New Junior Skill Stack (the real moat): Write tests before trusting outputs
- Use types, linters, and CI like seatbelts
- Reproduce bugs, isolate diffs, verify fixes
- Understand fundamentals: networking, databases, OS, security
- Ask better questions than the model can guess
Narendra N retweeted
I've been a developer for 10 years. Shipped 5+ apps using AI last year.
My STRONG advice for beginners building using Claude Code:
1. Claude Code works best when you give it one focused problem at a time.
Don't say: "Fix my app." Say: "This function returns undefined when the user is logged out. Here's the function. Here's the error. What's wrong?"
2. Use it to understand, not just to generate.
Most juniors just ship the code it produces. The ones who grow say: "Explain why you structured it this way."
You are not looking for an answer. You are looking for a mental model. One conversation where you understand the why beats 10 copy-pastes.
3. Sometimes, let it review YOUR code, not write it for you.
Write the function yourself first. Then ask Claude Code: "What is wrong with this? What would a senior engineer change?"
That's a code review on demand. That's the feedback loop that used to cost you a senior's calendar slot.
4. Use it to debug, not to guess.
Stop randomly changing things and re-running.
Paste the error. Paste the stack trace. Paste the relevant code block.
Ask: "Walk me through what is happening here, step by step."
Debugging is a skill. Use Claude Code to build it, not skip it.
5. The senior move - Plan well before executing
Use Claude Code to write the first draft of something unfamiliar - a new API integration, a new pattern, a new framework.
Then study that draft like it's a textbook. Your job is to understand what was generated, own it, extend it, and catch its mistakes.
The engineers who will win are not the ones who use AI the most. They are the ones who use it to compound their own understanding.
Claude Code isn't a shortcut. It's a multiplier - but only if you bring the fundamentals to the table.
Narendra N retweeted
Most people learn Kubernetes as a list of concepts. Pods. Deployments. Services. Ingress.
They memorize them without understanding why they exist.
>> You start with a Pod.
- A pod runs your container. Simple. Clean. Done.
- Until it crashes.
- Nobody restarts it. It is just gone.
In production, that is not acceptable.
>> So you use a Deployment.
- A Deployment watches your pods.
- One dies, and it creates another.
- You want 3 running, it keeps 3 running.
- You want to scale to 10; one command does it.
Pods were too fragile for production.
Deployments fixed that.
>> But now you have a new problem.
- Every pod gets a new IP when it restarts.
- You have 3 pods running your app.
- Another service needs to talk to them.
- Which IP do you use? They keep changing.
- You cannot hardcode them.
- You cannot track them at scale.
>> So you use a Service.
- A Service gives your app one stable IP address.
- It finds your pods using labels, not IPs.
- Pods die and come back with new IPs.
- The Service does not care.
- It always finds them.
- It also load balances.
- Traffic coming in gets distributed across all healthy pods automatically.
Pods had unstable IPs. Services fixed that.
>> But your app still needs to be accessible from the internet.
- So you use a LoadBalancer Service.
- This creates a real cloud load balancer.
- AWS ALB. Azure LB. GCP LB.
- Your app gets a public endpoint.
- Works perfectly. Until you have 10 services.
- Now you have 10 load balancers.
- Each one costs money every single month.
- Your cloud bill does not care that 6 of them handle almost no traffic.
LoadBalancer Services solved external access. But one per service does not scale.
>> So you use Ingress.
- One load balancer. All your services behind it.
- Ingress routes traffic based on rules.
- Request comes in for /api, goes to the API service.
- Request comes in for /dashboard, goes to the frontend service.
-One entry point. Smart routing. One cloud load balancer on your bill.
But Ingress is just a set of rules. Something has to execute those rules.
>> So you use an Ingress Controller.
- Nginx. Traefik. AWS Load Balancer Controller.
- These are the actual engines that read your Ingress rules and make the routing happen.
- Ingress without a controller is just a config file nobody reads.
To summarize it:
> Pod ran your app but had no resilience.
> Deployment gave it resilience.
> Service gave it a stable address and load balancing.
> LoadBalancer Service gave it external access.
> Ingress replaced 10 load balancers with one.
> Ingress Controller made the rules actually work.
Each concept exists because the previous one was not enough.
Narendra N retweeted
PostgreSQL versus MySQL
Built using the C language, PostgreSQL uses a process-based architecture.
You can think of it like a factory with a manager (Postmaster) coordinating specialized workers. Each connection gets its own process and shares a common memory pool. Background workers handle tasks like writing data, vacuuming, and logging independently.
MySQL takes a thread-based approach. Imagine a single multi-tasking brain.
It uses a layered design with one server handling multiple connections through threads. The magic happens using pluggable storage engines (such as InnoDB, MyISAM) that you can swap based on your needs.
Which database do you prefer?
Narendra N retweeted
If it’s real-time → WebSockets
If it’s scale → Kafka
If it’s simplicity → REST
If it’s chaos → GraphQL
If it’s AI → Python
If it’s infra → Go
If it’s logs → ElasticSearch
If it’s low-latency → Redis
If it’s high-availability → Postgres
If it’s streaming → Flink
If it’s low-level → C
If it’s high-performance → C++
If it’s enterprise → Java
Narendra N retweeted
If you want to become good at system design, then learn these 12 case studies (not kidding):
1 How ChatGPT Apps Work:
↳ newsletter.systemdesign.one/…
2 How YouTube Works:
↳ newsletter.systemdesign.one/…
3 How Google Docs Works:
↳ newsletter.systemdesign.one/…
4 How Kafka Works:
↳ newsletter.systemdesign.one/…
5 How WhatsApp Works:
↳ newsletter.systemdesign.one/…
6 How Airbnb Works:
↳ newsletter.systemdesign.one/…
7 How Spotify Works:
↳ newsletter.systemdesign.one/…
8 How Slack Works:
↳ systemdesign.one/slack-archi…
9 How Reddit Works:
↳ newsletter.systemdesign.one/…
10 How Bluesky Works:
↳ newsletter.systemdesign.one/…
11 How Twitter Timeline Works:
↳ newsletter.systemdesign.one/…
12 How Uber Computes ETA:
↳ newsletter.systemdesign.one/…
What else should make this list?
——
👋 PS - Want my System Design Playbook for FREE?
Join my newsletter with 200K+ software engineers right now:
→ newsletter.systemdesign.one/…
———
💾 Save this for later & RT to help other software engineers ace system design.
👤 Follow @systemdesignone + turn on notifications.
Most do it - Except that it might not just be tabs, as we also have IDE Plugins
In my view, Claude used to be (& probably still is) superior for Developers - Most use GPT, though
Gemini up'd their game with 3.0
Grok shows real difference in developer (to normal) mode
i saw a guy coding today
> tab 1 ChatGPT
> tab 2 Gemini
> tab 3 Claude
> tab 4 Grok
> tab 5 DeepSeek
he asked every AI the same exact question
patiently waited, then pasted each response into 5 different Python files
Hit run on all five
Pick the best one
Like a psychopath
It's me
Readers added context they thought people might want to know
This post was stolen word for word only to generate interactions
OP's post: x.com/i/status/19165…
Narendra N retweeted
If you want to become a better software engineer (in 2026),
read these 12 engineering blogs:
1. Meta Engineering
↳ engineering.fb.com
2. Netflix TechBlog
↳ netflixtechblog.com
3. AWS Architecture
↳ aws.amazon.com/blogs/archite…
4. Microsoft Engineering
↳ devblogs.microsoft.com/engin…
5. Google Research
↳ research.google/blog
6. Slack Engineering
↳ slack.engineering
7. Discord Engineering
↳ discord.com/category/enginee…
8. NVIDIA Developer
↳ developer.nvidia.com/blog
9. Stripe Engineering
↳ stripe.com/blog/engineering
10. Uber Engineering
↳ uber.com/en-AU/blog/perth/en…
11. Cloudflare Blog
↳ blog.cloudflare.com/tag/engi…
12. GitHub Engineering
↳ github.blog/engineering
What other blogs should be on this list?
--
👋 PS: Want my System Design Handbook for free?
Want my Architecture Patterns Playbook for free?
Join my newsletter with 26,501+ software engineers: lucode.co/luc-newsletter-lm1…
--
📌 Save for later.
♻️ Repost to help others grow.
➕ Follow Nikki Siapno + turn on notifications.
Narendra N retweeted
A must-read for LLM enthusiasts. Sebastian Raschka distills GPT-2 → GPT-OSS architecture beautifully 👏👏👏
Model Architecture from GPT-2 to GPT-OSS
magazine.sebastianraschka.co…
Narendra N retweeted
Here you go:
DeepSeek R1 is an AI model. An AI model is a bunch of a matrices with floating point numbers (referred to as weights) where you feed in an input (a sequence of characters embedded as a vector of floating point numbers) and get an output sequence.
DeepSeek is a mobile app (same name as the company) that lets you interact with that AI model through a chat interface. When you use their app, your data (prompts) go to their servers.
The company has also open sourced (basically uploaded all those matrices) the weights of the AI model for free use by anyone.
When you download those weights and bring it up yourself on your own server, you get to control the inference of the AI model and that way any user request sent to this new server doesn’t go to China as long as the servers are hosted in US.
The weights are just a bunch of numbers organized as matrices executed with sequential matrix multiplies - so no computation needs to leave the server in order to compute the next word in a sequence.
That way, another company can download the weights, host it on their servers, and let users interact with them in a chat frontend, and customize the AI model further to do more things like searching the web or using tools like code execution, wolfram, etc
