Mansi 👩‍💻

Senior Software Engineer | System Design | DSA | Interview prep | AI.
Follow me for update related to AI, tech, jobs, and software engineering career.

[Mansi 👩‍💻 @MansiCodez]

starting #30DaysOfAWS today

will share one concept daily —
things that actually matter in real projects

not just exam prep. real stuff. 🧵

follow along 👇

[Mansi 👩‍💻 @MansiCodez]

if you're also preparing for AWS Developer cert

these resources helped me week 1:

→ Stephane Maarek course (Udemy) — best in class
→ AWS official docs — underrated
→ building small projects — nothing beats hands on

starting week 2 tomorrow 🚀

what AWS topic confuses you the most?
drop in comments 👇

#30DaysOfAWS #AWS

[Mansi 👩‍💻 @MansiCodez]

circuit breaker = don't keep calling a dead service fail fast. recover faster. #Microservices

[Mansi 👩‍💻 @MansiCodez]

rate limiting = bouncer at the door controls how many requests per second #API

[Mansi 👩‍💻 @MansiCodez]

database sharding = split big table into smaller ones horizontal scaling for your data #Database

[Mansi 👩‍💻 @MansiCodez]

message queue = don't call me, i'll call you services talk without waiting #Kafka

[Mansi 👩‍💻 @MansiCodez]

CDN = copy of your content, closer to user latency drops. users happy. #CloudComputing

[Mansi 👩‍💻 @MansiCodez]

@thedatatreat @X Followed you!!

[Mansi 👩‍💻 @MansiCodez]

Hey @X algorithm
I'm looking to #connect with people interested in:
SaaS
Frontend
Backend
Full-stack
DevOps
App Development
AI / ML Data Science
LeetCode & DSA
System Design
Interview prep
Tech job/Career
Building in public
If that's you, let's connect 🤝

[Mansi 👩‍💻 @MansiCodez]

cache = memory for your app stop asking DB same question twice #Redis

[Mansi 👩‍💻 @MansiCodez]

@Rankit_Das_ @X Followed you

[Mansi 👩‍💻 @MansiCodez]

@danieldwalton Thank you 😊

[Mansi 👩‍💻 @MansiCodez]

starting #30DaysOfAWS today

will share one concept daily —
things that actually matter in real projects

not just exam prep. real stuff. 🧵

follow along 👇

[Mansi 👩‍💻 @MansiCodez]

load balancer = traffic cop for your servers one server shouldn't take all the hits #SystemDesign

[Mansi 👩‍💻 @MansiCodez]

@talkeen_in @X followed

[Mansi 👩‍💻 @MansiCodez]

@x twitter monetization requirements:

500 followers ✓ (almost)
500K impressions in 90 days ✓ (lol)

replies don't count as impressions btw

so all that engagement advice?
great for growth

terrible for monetization

x really said: talk less, post more 😅

[Mansi 👩‍💻 @MansiCodez]

@Kartikwt Thanks

[Mansi 👩‍💻 @MansiCodez]

honest week 1 reflection:

things I thought I knew but didn't:

→ IAM roles vs access keys (was doing it wrong)
→ S3 is way more than storage
→ DynamoDB is NOT just a NoSQL database
→ Lambda cold starts are real in production

7 days in. already learning a lot.

#30DaysOfAWS

[Mansi 👩‍💻 @MansiCodez]

HashMap is the most underrated topic in coding interviews.

It's not just "O(1) lookup." It's the backbone of caching, deduplication, DB indexes, and half the "easy" LeetCode problems that trip people up.

Everything you need on one card 👇 (bookmark this)

[Mansi 👩‍💻 @MansiCodez]

Alright, here's how I'd design it 👇

1. Gateway → WebSocket servers
Every client holds a persistent connection (not polling). On connect/disconnect, the gateway fires an event — that's your source of truth for "online" vs "offline."
2. Presence store → Redis
Key: user_id → {status, last_seen_timestamp}
Redis handles millions of reads/writes at sub-ms latency, and TTLs make "auto-offline after inactivity" trivial.

3. Fan-out → Pub/Sub (Redis Pub/Sub or Kafka)
When status changes, publish an event. Only fan it out to users who actually have that contact open/in view — this is the part most people get wrong. You don't broadcast to everyone, only to relevant "watchers."
4. Disconnect handling → Kafka / event queue
Sudden disconnects (phone dies, app killed) need a async worker to detect timeout and update Redis — don't do this synchronously in the request path.
5. Last seen persistence → Cassandra/DynamoDB
Redis is hot/in-memory; write last_seen periodically to durable storage so it survives restarts and can be queried historically.

The real trick isn't the components — it's not broadcasting to everyone. Fan-out-on-write to your actual audience (people with your chat open) is what keeps this from melting at 2B users.
What would you change? 👇

[Mansi 👩‍💻 @MansiCodez]

WhatsApp shows "last seen" for 2 billion users in real time.

No polling. No overload. Instant updates.

How would you design this system? 👇

(Drop your architecture in the replies — I'll share the best one)

[Mansi 👩‍💻 @MansiCodez]

@Alan1161816 @X sure

[Mansi 👩‍💻 @MansiCodez]

@SanketGangani17 @X followed you, your turn now.

[Mansi 👩‍💻 @MansiCodez]

@theignatpetrov @X followed you, your turn now

[Mansi 👩‍💻 @MansiCodez]

@rzasupreme @X why not. followed you, your turn now