Shivangi Sonie

Prod deploy happened. Everything looked clean. Service worked fine for 2 days. Suddenly on 3rd day, alerts start firing. But weird ones. 70% of requests working fine. 30% failing. Same endpoints. Same payload. Same everything. Junior team spent the whole day in the code. Rolled back the deploy. Still failing. Rolled forward. Same thing. Logs showed nothing. No exceptions. No timeouts. Just... requests disappearing. Here's the weird part: Some pods worked perfectly. Others failed on the same requests. Same image. Same config. Same cluster. What would you check?

2

DSA clears the first gate. System Design decides who actually gets hired — especially now. Recent interviews make one thing obvious: Companies are no longer impressed by candidates who only solve problems fast. They want to see how you think about real systems under constraints. After going through multiple interview rounds at top product companies, a clear pattern emerges: - A small set of system design problems keeps repeating - The questions change, but the thinking framework stays the same Most interviews revolve around variations of these core system design👇 🔹 Booking & Inventory Systems Avoiding double booking Handling cancellations, refunds, and race conditions 🔹 URL Shortening Platforms Generating unique links at scale Handling massive redirects with low latency 🔹 Chat & Messaging Systems Message ordering and delivery guarantees Offline users, retries, and real-time synchronization 🔹 Payment Workflows Idempotency and retry handling Failure recovery, security, and audit logs 🔹 Ad & Event Aggregation Systems High-volume data ingestion De-duplication and near real-time analytics 🔹 Caching Systems Fast reads and intelligent invalidation Balancing availability vs consistency 🔹 Social Feed & Timeline Systems Fanout strategies and ranking logic Privacy controls, pagination, and notifications 🔹 Rate Limiting Systems Protecting APIs from abuse Handling traffic bursts reliably 🔹 Webhook & Event Delivery Guaranteed delivery Retry strategies and failure handling 🔹 Trending & Analytics Systems Top-K problems over sliding time windows Memory-efficient real-time computation 🔹 Logistics & IoT Systems Order lifecycle tracking Hardware + software coordination 🔹 Ride Matching Systems Geo-based matching State management under heavy traffic 🔹 Media Hosting Platforms Storage, CDN, and transformations Abuse prevention and caching strategies If you zoom out, interviews aren’t testing tools or buzzwords. They’re testing decision-making under constraints. That’s why memorising answers no longer works. Most candidates ask: “Should I focus on DSA or System Design first?” The honest answer: 👉 DSA is non-negotiable. It’s the entry ticket. But DSA without the right patterns, order, and focus leads to random grinding. That’s what I’ve built for you: ✅ Notes that are concise and clear ✅ Most-asked questions/topic ✅ Real patterns + approaches to master them 👉 Grab the DSA Guide → https://lnkd.in/d8fbNtNv (300+ devs are already using) 𝐅𝐨𝐫 𝐌𝐨𝐫𝐞 𝐃𝐞𝐯 𝐈𝐧𝐬𝐢𝐠𝐡𝐭𝐬 𝐉𝐨𝐢𝐧 𝐌𝐲 𝐂𝐨𝐦𝐦𝐮𝐧𝐢𝐭𝐲 : Telegram - https://lnkd.in/d_PjD86B Whatsapp - https://lnkd.in/dvk8prj5 Built for devs who want to crack interviews — not just solve problems.

59

6 Comments

Avoiding if statements in loops The high cost of if statements goes way down to the hardware level. Each if statement creates a branching instruction. The problem is that unlike continuous code execution, which is like a highway road, the branching execution is like a crossroad with a traffic light. Thankfully, processors have a way to try to predict where the code should go next and if they guess correctly - the speed is almost the same. It's comparable, if you see the green light from a distance and don't slow down at the crossroad. The real problems begin if the guess it's wrong. It's called a branch misprediction. It's like breaking to a full stop and going back to the correct turn on the crossroad. Now imagine that this is happening hundreds of times where sometimes we predict correctly and the other times we must fully stop and choose the other correct way. This basically what it means to implement an if statement inside of a loop. What makes the situation worse, is when there are multiple nested loops with a complex if/else if/else statement inside. This way, we will need to try different branches before we get to the correct one. And as we do that multiple times the slowdown will be critical. The solution is quite simple - move if statements outside of loops. This way, the possible mispredictions will only happen several times, not hundreds of thousands. If there is no way to avoid if statements that should be executed multiple times, there is another thing that should be taken into an account. In general, most compilers will order the produced machine code roughly the way you ordered it in your code by default. So, you should order your branches in the order of decreasing likelihood to get the best branch prediction from a "first encounter".

16

1 Comment

🚀 Scaling Isn’t About Bigger Servers. It’s About Smarter Architecture. Recently, I watched a talk by Kailash Nadh, CTO of Zerodha, on how they handle millions of records in PostgreSQL without breaking production systems. One idea stood out: ❌ Don’t run heavy reports directly on your transactional database. ✅ Offload heavy computation to asynchronous background workers. When users request large reports (like multi-year trading statements), instead of executing massive SQL queries immediately: • The request is queued • Background workers process data in chunks • Results are stored separately • The main system remains fast and stable This reinforces a powerful engineering principle: 💡 Databases are excellent at storing and retrieving data. They are not designed to be heavy compute engines. Key Takeaways: ✔ Separate transactional workload (OLTP) from heavy reporting ✔ Use async processing for expensive tasks ✔ Break large jobs into smaller chunks ✔ Design for system stability, not just feature completion As someone exploring distributed systems, Saga patterns, and microservices architecture — this was a strong reminder: Scalability is more about architectural decisions than hardware upgrades. Curious — how do you handle heavy reporting in your systems? #BackendDevelopment #SystemDesign #Microservices #PostgreSQL #DistributedSystems #Scalability

4

System Design is not something you only worry about once you hit Senior SDE, trust me 👇 Because in reality, System Design shows up much earlier, and it’s a very imp. filter for any roles beyond SDE-2/3. That’s why I’m sharing this “System Design Made Easy” doc 📄 Big thanks to Bosscoder Academy for curating this structured guide! 🙌 📌 Check their programs here → bcalinks.com/Rr38sAy Here’s what you’ll find inside the doc: → Core concepts explained step by step (scalability, caching, load balancing, CAP theorem, databases) → Breakdowns of real-world system architectures → A 60-day roadmap designed to fit alongside a full-time job If you’re a mid-to-senior SDE prepping for your next role, this doc can save you weeks of scattered prep and give you a roadmap that actually works. And if you’re serious about growing into Sr. SDE roles at top PBCs, Bosscoder Academy’s EVOLVE program for SDEs with 6+ YOE is built exactly for this. They’ve already helped 2200+ engineers crack top PBC roles, maybe you’re next? #systemdesign #seniorsde #techinterviews #growth #jobswitch #collab

142

8 Comments

🧠💥 Hey, it’s 4th Nov 2025 — and OpenAI just flipped the switch! ChatGPT Go is now FREE for a whole year across India 🇮🇳 Yep, you read that right the mid-tier plan (usually ₹399/month) is now free for 12 months. And this applies to both new and existing users! 🎉 If you already had ChatGPT Go, no worries your subscription automatically extends for a full year at no cost. 💡 What Exactly Is ChatGPT Go? It’s basically an affordable version of "ChatGPT Plus" built to give you better features without paying the full price. Here’s what you get 👇 ⚡ GPT-5 access (latest, faster model) 📁 Upload PDFs, Docs & Sheets for AI analysis 🖼️ Generate & edit images 🧠 Personalised memory for smarter chats 💬 Longer conversations + higher limits 🚀 Priority access even during peak hours 🪄 How To Activate the Free Plan 👇 1️⃣ Go to chat.openai.com or open the ChatGPT app 2️⃣ Log in or sign up (Google login works too) 3️⃣ Tap your profile → Settings → Subscription 4️⃣ Select ChatGPT Go —> it’ll show “Free for 1 Year” offer for India 5️⃣ Add a valid payment method (Card or UPI) 6️⃣ Verify via email or OTP if asked Once done ✅ your account upgrades instantly — free for 12 months! 💳 OpenAI requires a card or UPI mainly for two reasons: 1️⃣ To verify your region & prevent spam 2️⃣ To enable auto-billing after 12 months 👉 Tip: if you only want the free year, set a reminder to turn off auto-renewal before it ends, so you don’t get charged later. ❌ What’s Not Included : No API access No GPT-4 Turbo or GPT-4o No external connectors or Sora video tool No manual control over GPT-5 “thinking mode” 🔥 Honestly, this is a great deal — a full year of advanced AI features for ₹0. I’ve already activated mine — if you haven’t yet, just takes 2 minutes. #OpenAI #ChatGPTGo #ArtificialIntelligence #AIIndia #ProductivityTools #ChatGPT #AIRevolution #FutureOfWork #TechNews #IndiaTech #GPT5 #AIUpdates #Innovation #AIFreeAccess

3