Fat FIRE devs

My friend and I are both software engineers. We are same age. Have similar skill level. We both write backend code. And we both sit in front of a laptop for 8 to 10 hours a day. But one small difference is slowly making his life very different from mine. He earns in dollars. I earn in rupees. At first, I did not think much of it. I used to think, salary is salary. Work is work. Code is code. But over a few years, I understood what was really happening. Every month, he gets paid in USD. When he converts that money to INR, he gets a stronger payout. And when the rupee gets weaker against the dollar, his income in rupee terms rises even if his dollar salary stays the same. That is the part many software engineers in India do not fully understand. When people say: "The rupee fell against the dollar" It simply means you now need more rupees to buy 1 dollar. Example: If 1 USD = ₹80 and later 1 USD = ₹88 that means the rupee has weakened. Earlier, $1,000 = ₹80,000 Now, $1,000 = ₹88,000 Same work. Same client. Same dollar payment. But now the person earning in dollars gets ₹8,000 more just because of currency movement. This is why my friend keeps getting richer in rupee terms without even asking for a raise. Now look at the other side. If you live in India and earn in rupees, a weaker rupee quietly makes many things more expensive over time: - imported gadgets - foreign software tools - international travel - overseas education - dollar-priced subscriptions - even some fuel and inflation-linked costs So while your salary may go up 8 to 10 percent, your purchasing power may not improve much. This is the trap. Many Indian software engineers think: "I got a raise, I am doing well" But if the rupee keeps weakening and inflation keeps rising, you are running hard just to stay in the same place. My friend understood this early. He did not become a genius overnight. He just did 3 simple things: 1) learned skills global companies pay for 2) started working with clients who pay in dollars saved 3) and invested instead of inflating lifestyle That is it. No flashy startup. No rich dad. No lucky crypto trade. Just one decision: stop thinking like a local employee, start building like a global engineer. This is why earning in dollars is not just about a bigger salary. It is about: better currency better savings power better global optionality and faster wealth creation if you stay disciplined Indian software engineers need to understand this clearly: - earning in rupees can make you comfortable - earning in dollars can accelerate wealth - building equity or business can change your life Salary matters. But the currency of that salary matters too. If your goal is real wealth, not just monthly survival, you cannot ignore the dollar vs rupee game.

Uber India compensation (rough ranges I’ve seen/heard, base + bonus + RSUs, in ₹ LPA). 1. SDE-1 (L3): 35 to 55 2. SDE-2 (L4): 60 to 95 3. Senior (L5): 1.0 to 1.6 Cr 4. Staff (L6): 1.6 to 2.5 Cr 5. Sr Staff (L7): 2.5 to 3.5+ Cr --- Uber interview experience 1. Recruiter screen (20 min) - Confirmed role level, location, TC band. - Asked for 2 projects where I moved a metric (latency, cost, reliability). - Tip: say numbers. “p95 900ms → 140ms” beats “improved performance”. 2. Coding round 1 (60 min) - Medium DS problem. Clean solution was more important than tricks. - They cared about: constraints first, tests, then code. - I narrated edge cases: empty input, duplicates, overflow, time limits. 3. Coding round 2 (60 min) - Slightly harder. Required good data structure choice. - I wrote a brute force quickly, then optimized. - Biggest signal: can you debug calmly when your first approach breaks? 4. System design (75 min) - “Design a ride matching / dispatch-like system” (variant). - What they pushed on: 1. Peak QPS and hotspots (city centers, events). 2. Geospatial indexing (H3/S2, grid bucketing). 3. Real-time updates (driver location stream), stale data, and consistency. 4. Matching strategy (nearest vs ETA), fairness, cancellation handling. 5. Failure modes: region outage, queue backlog, thundering herd. - I got points for explicitly calling out idempotency + retries + timeouts. 5. Hiring manager (45 min) - Less “tell me about yourself”, more “tell me what you shipped”. - Deep dive on one incident: what broke, how we detected it (metrics/alerts), and what changed after. What I’d do again: 1. Start every round with constraints + tradeoffs. 2. Use concrete numbers. 3. Treat interviews like production: observability, failure modes, rollback.

Your Redis server is at 100% CPU. Profiling shows KEYS * command being called repeatedly. What's wrong and how do you fix it?

Wild times! And people are at war. War against each other. War against their own focus. War against patience. War against the truth they already know but keep avoiding. We are living in a time where tools like GPT can help us push science forward, maybe even help in things as big as curing cancer. But are we really using this technology to uplift the world the way we should? I do not think so. The real problem is not always the tech. It is us. Humans are often too distracted, too selfish, too busy chasing personal gain to ask a bigger question: How do we use this to reduce suffering, spread knowledge, and move humanity forward? Technology alone does not make the world better. Intent does. We need to learn how to give, not just take. Only then this kind of technology will truly help us evolve and build something meaningful for the future.

My friend moved to Rajkot for a ₹72k/month role at a "quantum AI infrastructure startup" based out of an industrial lane behind a tyre shop. Yes, quantum physics. In Rajkot. Naturally Bangalore people had jokes ready. - Mechanical engineering graduate - Switched from support to data to AI - Fourth job in 3 years - Moved there for "career growth" Salary: ₹72,000 per month In Bangalore, people smirked. "Bro left the tech capital for a physics experiment," they said. "Next he will build AGI from a paan shop." But 6 months later the joke started looking expensive. - Rent for a clean 2BHK near Kalawad Road: ₹12k - Home style tiffin + groceries: ₹7k - Commute by scooter: ₹2k Tea, fafda, khaman, random snacks: ₹2.5k Weekend spending: somehow still under ₹3k And then comes the hidden arbitrage no one puts in CTC comparison posts. - No ₹1200 sushi dinners to "network" - No ₹250 coffees to discuss burnout - No pressure to live in a matchbox for the sake of pin code prestige - No spending half your life in traffic just to prove you work in tech - No startup cosplay culture where everybody looks rich and is secretly one layoff away from panic By month end he was saving ₹40k+ without doing monk mode. That is when he understood something funny about Indian salaries. ₹1.8L in a metro can still make you feel broke. ₹72k in a smaller city can make you feel like an adult who has his life together. Big city people sell you one dream: high CTC, low peace. Small town people accidentally discover another one: decent pay, low noise, real savings. Call it what you want. Quantum computing maybe still hard. But this math is not. --- The story is fake but you get the idea ;)

Most system design posts online are just boxes and arrows. A real one should have: 1. Problem + numbers (RPS, p95, data size, SLOs). 2. API + data model (tables, keys, indexes, retention). 3. Critical path (what happens on the hot request). 4. Failure modes (timeouts, retries, idempotency, DLQ). 5. Tradeoffs (consistency vs availability, cost vs latency). 6. Rollout plan (feature flags, backfill, migration, kill switch). 7. Observability (logs, metrics, traces, dashboards, alerts).

8 Ways for a Software Developer to Grow and Stay Relevant 1. Code every week 2. Learn one core concept deeply 3. Build small side projects 4. Read good code from real repos 5. Get better at system design 6. Write about what you learn 7. Improve communication 8. Track your weak areas and fix them Bonus: stay updated with AI trends, especially coding tools and agents, because they are becoming part of everyday engineering work

When backend work starts feeling too repetitive, I like switching to a completely different kind of problem. Something visual. Something from the real world. A while back I got obsessed with building small simulations just for fun. And not for money or interviews. Just because it made programming feel alive again. I would pick random things and try to model them from scratch. Price movement as particles. Order books as flowing systems. Network packets like traffic. Even simple market behavior starts looking very different when you try to simulate it instead of just reading about it. You stop thinking like a CRUD developer for a while and start thinking in terms of motion, constraints, feedback loops, probability, emergent behavior. And honestly, that is where programming starts feeling fun again. These days it is even easier. You have better languages, better libraries, better visual tools. You can focus on the core logic instead of fighting the setup for 3 days. Every software engineer should do this once in a while. Build something slightly useless. Slightly weird. Slightly beautiful. It reminds you that coding is not just tickets, APIs, and fixing late night production bugs. Sometimes the best way to stay sharp is to make something that teaches you how things work. Always remember that you are down this career path because you enjoy building things.

Reviewed an anonymized backend resume from a 1.5 to 2 year experience engineer building ad-tech + RAG products. The good: Strong impact signals. Numbers like 3,000 RPS steady, burst traffic handling, latency cuts, event scale, dashboard latency, and actual architecture words like retries, backpressure, graceful degradation make the resume feel real. Projects also show ownership instead of just tutorial work. The bad: It tries too hard to sound senior in some places. Claims like architected fault-tolerant systems, scaled to massive burst traffic, designed multiple persistence layers, built analytics pipelines, all in a short career span can make recruiters doubt what was personally owned vs team effort. Good resumes should be impressive, but also believable. The ugly: A few bullets feel like system design interview answers pasted into a resume. Recruiters do not just want big words, they want clarity. What exactly did you build, what part did you own, what changed because of your work, and what tradeoff did you handle. Also too many tools are listed. Skill sections become weaker when they look like keyword stuffing. What I would improve: 1. Split ownership clearly: built, led, contributed to, optimized. 2. Add tech stack per experience bullet where relevant. 3. Reduce inflated architecture wording and make it more concrete. 4. Keep only strongest 5 to 6 skills that match target backend roles. 5. Put best proof of work first, not every possible achievement. 6. Add one line on team size or cross-functional work. 7. For projects, mention users, revenue, waitlist, or production usage more clearly. 8. Remove vanity links and anything personal before posting online. Overall: very solid base resume. Good enough to get shortlisted for backend roles, but with better honesty and clearer ownership, it can look much stronger. Big lesson: numbers get attention, but believable ownership matters more!

Be James Harris Simons > earned his B.S. from MIT at age 20 > got his Ph.D. from UC Berkeley at just 23 > worked for the NSA and the IDA during cold war > became the chair of the math department at Stony Brook University at age 30 > co-developed the Chern–Simons form (foundational to String Theory and quantum physics) > was awarded the Oswald Veblen Prize in Geometry, the highest honor in the field > left academia to start a hedge fund, Monemetrics at age 40 > rebranded his firm to Renaissance Technologies - hired physicist, math PhDs > founded non-profits to improve math education in US and for autism too > named "the world’s smartest billionaire" by the Financial Times NET WORTH: $31 billion Jim Simons passed away on May 10, 2024, at the age of 86. He often cited "Five Guiding Principles" for his life: 1. Be original: Do something new; don't run with the pack. 2. Collaborate: Surround yourself with the smartest people you can find. 3. Be guided by beauty: A mathematical proof or a well-run business can be beautiful. 4. Don’t give up: Stick with it until it works. 5. Hope for good luck: And recognize it when it happens. youtube.com/watch?v=CTQcLi…

No bullshit system design guide for backend engineers who want to reach Staff level. Devs stay stuck at ₹10 to 25 LPA cause they know frameworks, but not systems. Meanwhile Staff Engineers are often paid: India: ₹40L to ₹1Cr+ Remote: $120k to $250k+ So if you cannot explain these clearly, you are not ready for senior backend roles, let alone Staff. 1. Load Balancing 2. SQL vs NoSQL 3. Idempotency 4. Message Queues 5. CAP Theorem 6. APIs 7. Batch vs Stream Processing 8. Caching Strategies 9. Webhooks 10. Availability 11. Data Sharding and Partitioning 12. Bloom Filters 13. Stateful vs Stateless Architecture 14. Algorithms in Distributed Systems 15. API Gateways 16. Proxy vs Reverse Proxy 17. Sharding 18. Long Polling vs WebSockets 19. Consistent Hashing 20. gRPC, tRPC, GraphQL, or REST 21. Caching 22. Scaling 23. Cache Eviction Policies 24. Databases in System Design 25. JWTs 26. Services in System Design 27. Concurrency vs Parallelism 28. CDC 29. ACID Transactions 30. CDN 31. Sync vs Async 32. Rate Limiting Algorithms 33. REST 34. gRPC vs REST tradeoffs 35. Fault Tolerance Truth is, Staff level is not about writing more code. It is about knowing where systems break, why they break, and how to design so they keep making money even when traffic, failures, and complexity go up.

I have a friend who spent 6 years giving everything to his job. Late nights. Weekend deploys. Production issues at 2am. Helping the company hit targets he would never really benefit from. Every year the same story: small raise, bigger responsibilities, new title maybe, but same invisible ceiling on income and freedom. One day he got tired of realizing that his best skills were being rented out cheaply while the company kept all the upside. So he started small. Took one client on the side. Built a tiny backend automation project for them. Then another client came through referral. Then one more. That little side hustle slowly became an agency. Now he still works hard, maybe harder than before honestly, but there is a huge difference: every process he improves, every client he helps, every system he builds adds equity to his own life, not just to some faceless corporation's quarterly report. I think this shift will become very common. Not everybody needs to quit tomorrow and start a company, but more people should at least try to own a piece of what they build. Uncapped ownership changes how you see your time, your energy and your future. Fr.

You deploy a change that slowly leaks memory. Monitoring shows memory growing 1% per hour. No alerts fire because change is gradual. How will you detect slow degradation? [Real incident at Uber]

Productivity habits that actually work for me as a Senior Software Eng: 1. Calendar the hard stuff (90–120 min). No Slack. One outcome: “ship X” or “debug Y”, not “work on Z”. 2. Write before coding. 10–15 lines in a doc: goal, non-goals, risks, rollout, metrics. Saves days of rework. 3. Kill meeting debt. Decline by default. If I accept, I ask for an agenda + decision needed. 25/50 min only. 4. Keep a “hot list” of 3 things. If it’s not on the list, it’s background noise. I review it every morning. 5. Debug with proper feedback and receipts. Repro steps, logs, traces, one graph. Then fix root cause. No “seems fine now” merges. So time is not your bottleneck. Context switching is.

Go concurrency is easy to demo, hard to run in prod. Some tips to write concurrent go code running in prod. 1. Goroutines are cheap until you spawn 200k and your “fast” service turns into a scheduler benchmark. 2. Channels don’t save you from backpressure. If nobody is draining, you’re just queuing pain. 3. `context` is not optional. Every goroutine needs a cancellation path, or you’re building leaks with style. 4. Start with a worker pool + bounded queues. Measure p95, goroutines count, and allocs before adding cleverness. So treat concurrency like capacity planning, not just syntax!

The goal is to earn capital and deploy that capital to generate more capital. Reasons for taking the trade today - Max pain: 79k - OI gap on the upside - High TFs RSIs making higher lows - Sensex at support - Got high premium because of gap down open Will explain those in the subsequent posts #VerifiedBySensibull verified.sensibull.com/ps/guiding-jui…

@Rushu_Tushu I am also blocked, guess what i counter blocked him, I am not here to make friends, but work with and learn from the best! If someone can't take an opinion, they are simply not worth it.

If you only earn and don't invest, you're essentially losing money due to inflation every year.

We often talk about idempotency but overlook the need for a good data store to persistently hold the idempotency key for deduplication. Have you ever thought about why? 1. The data store must be fast enough. If the idempotency check takes 300 ms (DB round trip + locks), retries pile up. 2. If you can’t retain the key longer than the retry window, you’re just guessing. The duplicate may arrive after 2 minutes. 3. If writing the key isn’t atomic with the side effect, you get the worst bug: “charged once, recorded zero” or “recorded twice, shipped once”. A fast store (Redis, DynamoDB, Postgres with proper unique index) isn’t just about speed hype. It’s about making the “check + claim” operation both cheap and correct.

Your microservices communicate over HTTP. At high load, you see "connection refused" errors even though services are running. What's the network issue?