abhijitWebDev

[abhijitWebDev @abhijitWebDev]

Just finished a system design session on Caching & Distributed Systems 🔥 — here's the full breakdown in a thread 🧵

1/ Why cache at all?
If a computation takes 30-40s, you cache the result. Next time someone asks — it's instant. That's the entire idea.

2/ Redis vs Postgres
Redis = RAM → blazing fast, limited size
Postgres = Disk → slower, persistent, large
Pick based on your use case — not hype.

3/ Cache HIT vs MISS
HIT → serve from Redis ✅
MISS → fetch from DB → update Redis → serve
This pattern = Read-Through Cache

4/ TTL is your best friend
Set expiry on every cache key.
24hrs for slow-changing data.
5min for real-time data.
No TTL = stale data forever 💀

5/ Eviction Policy: LRU
When storage is full, kick out the Least Recently Used item.
This is how Redis stays lean under pressure.

6/ Consistent Hashing 🔄
Distributed cache across 3+ nodes?
Use a hash ring.
hash(key) % n → determines the partition
Add/remove a node? Only a fraction of keys move. Not all.

7/ Infrastructure ladder:
Data Center → Hypervisor (Proxmox)
→ VMs → VPS → Cloud (AWS/Azure/GCP)
More abstraction = less control but less headache.
Thank you @piyushgarg_dev sir for the awesome session and we missed you today @Hiteshdotcom sir and thank you all the TA's @nirudhuuu @ChaiCodeHQ

#SystemDesign #Redis #Caching #DistributedSystems #ChaiCode #BackendEngineering

[Tushar Nebhnani @TusharNebhnani_]

Money can buy you GPU but not optimisation.

@ChaiCodeHQ @Hiteshdotcom @piyushgarg_dev @nirudhuuu @BlazeisCoding

[Hitesh Choudhary @Hiteshdotcom]

Interesting kubernetes story

[abhijitWebDev @abhijitWebDev]

@piyushgarg_dev Congratulations Sir!!

[Dhruv Sood @dhruvsood05]

Just wrapped up another GenAI class by @piyushgarg_dev

Today we went beyond basic RAG and explored how production-ready AI systems actually retrieve and reason over information.

Learning from today's class:

→ Query Translation
→ Step-Back Prompting
→ Query Routing
→ HyDE (Hypothetical Document Embeddings)
→ Reciprocal Rank Fusion (RRF)
→ Input & Output Guardrails
→ PII Detection
→ Advanced RAG Pipelines

@ChaiCodeHQ @Hiteshdotcom @nirudhuuu @BlazeisCoding @yntpdotme @devwithjay @surajtwt_

#chaicode #RAG #AdvancedRAG

[abhijitWebDev @abhijitWebDev]

RAG doesn't kill hallucinations — it just moves where they happen.
Poor retrieval, bad chunking, context limits, stale knowledge bases can all quietly break a RAG pipeline.
Wrote up where RAG actually fails (and when to skip it):
https://t.co/Om2K9FmAKG
@ChaiCodeHQ
#GenAI #RAG #LLM

[Tushar Nebhnani @TusharNebhnani_]

Building logic for DSA is Problem solving and not able to write that logic in code is programming problem. Both are separate issues.

The moment you realise this, DSA becomes easy.

@Hiteshdotcom

@PriyaBhatiaDS The best course of internet, everyone just jumps into the code. She taught the maths behind it.

[Anirudh @nirudhuuu]

unforgettable milestone for Indian engineering 🇮🇳❤️

[abhijitWebDev @abhijitWebDev]

Ever tried listing every skill installed across your Claude Code / Cursor / Copilot setup? Yeah, neither could I.
Skill-discovery, built by Surenode — one command, runs locally, zero config. Scans your agents' skill folders + instruction files, fingerprints them, flags suspicious patterns (env dumps, hidden payloads, prompt injection), redacts secrets before anything leaves your machine.
git clone https://t.co/YKKe15m4q8
cd skill-discovery && pip install .
skill-discovery --dry-run
Nothing leaves your machine unless you point it at a server. Open source, and it's the inventory layer behind Surenode's KLIN governance platform for teams.
🔗 https://t.co/YKKe15m4q8

[Suraj Jha @surajtwt_]

DSA in Java becomes much easier after practicing it visually on this amazing website by my @Hiteshdotcom.

[Atharv Dange @atharvdangedev]

Never forget that you are the main data center.
Drink water, and consume as much literature as possible.

[Tushar Nebhnani @TusharNebhnani_]

We have started building chatGPT clone.

@ChaiCodeHQ @Hiteshdotcom @piyushgarg_dev @surajtwt_ @nirudhuuu @BlazeisCoding

[Janhvi Deshmukh @janhvi_1508]

So true.....@Hiteshdotcom Sir & @nirudhuuu Sir please bring it back 🤌🏻🙌🏻

[abhijitWebDev @abhijitWebDev]

@sameerbhagtani @ChaiCodeHQ Congratulations !!!

[Piyush Garg @piyushgarg_dev]

🥹🥹

[Chai Aur Code @ChaiCodeHQ]

Milte hain phir aaj class mein. Civil uniform mein aana, aur snacks khud lekar aana ❤️

Thodi baatein, thodi yaadein... saath lekar aakhri manzil tak pahunchenge

This is not a full stop... it's just a comma

[abhijitWebDev @abhijitWebDev]

Class-01 of GenAi Done
Spent time breaking down how LLMs actually work, from scratch 🧠
1️⃣ Tokenisation → text becomes numbers
2️⃣ Embeddings → numbers become meaning (vectors that capture relationships, e.g. Eiffel Tower ↔ Paris)
3️⃣ Positional Encoding → order matters
4️⃣ Self-Attention → tokens "talk" to each other (ICICI Bank vs River Bank — context decides meaning)
5️⃣ Transformer (encoder-decoder) → training (backprop) vs inference

#GenAI #LLM #AIEngineering
@ChaiCodeHQ @Hiteshdotcom and @piyushgarg_dev sir for an awesome session.

[Dipali Sharma @sharmadipalidev]

Shipped my 1st full-stack project🌻
Command Inbox, for the Corsair Hackathon
https://t.co/488DyiBLr1
First time buying a domain, connecting a DB and surviving deployment failures...

@Hiteshdotcom @piyushgarg_dev @yntpdotme @nirudhuuu @BlazeisCoding @ChaiCodeHQ

[Piyush Garg @piyushgarg_dev]

Tonight, Things gonna get spicy 🥵 Because our GenAI cohort is starting 🫰🏻

[abhijitWebDev @abhijitWebDev]

@GithubProjects https://t.co/Ir1iEBgyDw

[Anirudh @nirudhuuu]

If you're excited about building AI applications and agents, you'll love our upcoming GenAI JS Cohort starting on June 29th

We'll go deep into LLMs, RAG, AI Agents, MCP, embeddings, vector databases, and production-grade AI systems.

Join us: https://t.co/MOOr64M7bg

See you in the class 🧡