Great paycheck, crap personal life? There's a life where professional success and fulfillment DO MIX. Let's craft it together💥 | Life & business coaching | Rich & Happy Club UAE | Founder The French Bijouterie 💎

Most people think the hardest part of success is building it.
It’s not.

The hardest part is knowing what to do once you’ve already won.

I work with founders, executives, and high performers who are doing well —
but feel a growing gap between the life they’ve built and the life they want to live next.

They’re not lost.
They’re at a crossroads.

The problem isn’t motivation or discipline.
It’s decision-making — when every option looks good, but each one creates a different life.

Before founding Changing Minds, I became a CFO at 26.
On paper, it was a win.
In reality, I wasn’t enjoying it — and I didn’t have a clear way to decide what came next.

That’s why I built Three-Dimensional Wealth™ Mentoring.

A decision-making process built on three pillars:

Meaning — what gives direction, energy, and long-term satisfaction
Mastery — how you regulate yourself, think clearly, and lead under pressure
Money — how you build and protect real wealth

Using concrete frameworks, tools, and structured introspection I use with my clients one-on-one, the goal is simple:

- make decisions faster
- get clarity on your next move
- feel confident committing to it
- be mentally ready to execute
- build wealth without sacrificing the life it creates

That’s the work I do.

This isn’t coaching for balance.
It’s for people who want more — and want to choose wisely.


If this reflects where you are and the next decision matters,
book a first call and get complimentary access to the Mindset Factor Assessment so we can get clear from day one.

-> https://calendly.com/l-lachaux/introduction-call

Lucie Lachaux

5 reasons smart people get confused about their next level
 
1️⃣ They have too many good options.  
Nothing is broken, so choosing feels risky.
 
2️⃣ What got them here still works.  
It pays the bills, looks fine, and keeps life stable — just not satisfying anymore.
 
3️⃣ Every option feels expensive.  
Change the job → risk income.  
Start the business → risk comfort. 
 Slow down → risk image.
 
4️⃣ They keep waiting for things to “slow down.”  
But life rarely does at this stage.
 
5️⃣ They’re mentally tired.  
Even when they have time, they don’t have the energy to think clearly.
 
Feeling confused here doesn’t mean you’re failing.  
It means you’ve reached a decision point.
 
P.S. What decision keeps coming back to your mind?

“Great Paycheck, Crap Life?” | Helping high-performing leaders facing pivotal life & career decisions — without losing what they’ve built | Creator of 3D Wealth™

This week, I felt completely flat.
 
No motivation. No direction. Just bored of my own life.
 
Here are 3 ways I wouldn’t recommend handling it:
 
1 - Push through with more work  
You think productivity will fix it. It won’t. You’re not tired from lack of output. You’re tired from lack of meaning.

2 - Distract yourself with plans  
Booking trips, starting projects, saying yes to everything. 
This just adds noise. It doesn’t solve the internal friction.

3 - Wait for it to pass  
You tell yourself it’s temporary. But if this feeling keeps showing up, it’s not a mood. It’s a signal.
 
This is how I’d do it instead:
 
1. Ask what’s missing — not what’s wrong.
 
• Money: Is this actually a money or growth problem?  
• Meaning: Do I still care about what I’m building?  
• Mastery: Are my thoughts and emotions running me?
 
2. Protect the silence. Don’t fill the void with people or plans. Sit with it. The answer is already there.
   
3. Make one small decision that feels true.  Not big. Not bold. Just honest.
 
Remember:
 
Flatness isn’t failure. It’s your system telling you something needs to shift.
 
👉 The next edition of 𝗧𝗛𝗘 𝗕𝗥𝗜𝗘𝗙𝗜𝗡𝗚 is about rock bottom — and the advantage no one talks about.
Subscribe here : https://lnkd.in/dYgGPSGX

2 harsh truths most successful people don’t like hearing:
 
1- Working harder won’t fix a life that’s slightly misaligned.
 
You can keep hitting targets.  
Keep getting promoted.  
Keep being “the reliable one.”
 
And still feel off.
 
Because effort isn’t the issue.  
Direction is.
 
2- Intelligence becomes a trap when you use it to justify staying the same.
 
You can explain why now isn’t the right time.  
Why it’s “strategic” to wait.  
Why this is just a phase.
 
And five years pass.
 
Not because you weren’t capable.  
Because you let your blind spots run the show.
 
At a certain level, growth isn’t about doing more.
It’s about resetting:
 
• how you evaluate decisions  
• what you tolerate  
• what you believe you deserve
 
That’s the real upgrade.
 
Not a new job.  
Not a new title.  
A new standard for how you operate.
 
If you feel the gap between what you built and what you actually want next — that gap doesn’t close with more effort.
 
It closes when you reset how you think and decide.

👉 Every month, I work privately with a small number of executives and founders on going back to the foundation ( how they think, decide, and define success ) so they get clear on their next move and move forward with confidence.
 
DM 𝟯𝗗 𝗪𝗘𝗔𝗟𝗧𝗛 if you’re ready to reset instead of repeat.

The most expensive mistake smart leaders make?
 
Staying too long in something that already works.
 
On paper, everything is fine.
 
Good money.  
Strong team.  
Solid reputation.
 
Nothing is broken.
 
But you keep thinking about the next move.
 
Not because you’re unhappy.  
Because you know you’re capable of more.
 
This is where leaders get stuck.
 
They don’t lack strategy.  
They lack a decision.
 
So they optimize what exists.  
Improve margins.  
Restructure teams.  
Add initiatives.
 
But they avoid the real question:
 
Is this still the right vehicle for where I’m going?
 
At a certain level, growth isn’t about doing more inside the same model. 
It’s about deciding whether the model itself needs to change.
 
And that’s uncomfortable.
Because the current version works.
But working isn’t the same as expanding.

P.S. Are you still building — or just maintaining what already works?

I help in skyrocketing 🚀 your career & finances | Engineering Manager @ Amazon | Ex GS | IIMK | Career coach | Personal finance coach

4 years before I got demoted from leading a team of 15 engineers to 1 engineer.

I doubted myself, if I am good at leading teams or not.

My managers, colleagues (𝘪𝘯𝘥𝘪𝘳𝘦𝘤𝘵𝘭𝘺) told me to stop leading the team.

Then one day without any agenda, I called my ex-manager, who then had moved to Amazon.

We spoke for an hour. He reminded me of my work and abilities when he used to be my manager.

He told me what matters in career and how I was in a wrong place at a wrong time

We then met again, but this time with agenda of exactly what I needed to do, to crack Amazon interview.

I applied, and cracked the interview in the very first attempt.

And after joining Amazon, I already got promoted once.

𝘛𝘩𝘢𝘵 𝘶𝘯𝘪𝘯𝘵𝘦𝘯𝘥𝘦𝘥 𝘤𝘢𝘭𝘭 𝘸𝘢𝘴 𝘵𝘩𝘦 𝘵𝘶𝘳𝘯𝘪𝘯𝘨 𝘱𝘰𝘪𝘯𝘵 𝘪𝘯 𝘮𝘺 𝘤𝘢𝘳𝘦𝘦𝘳.

Which taught me 3 very important lessons.

# 1
You are given one life.
Don't spoil it because of being in wrong place.

# 2
Be closer to those who motivate & mentor you.
Avoid everyone else.

# 3
Learning is the only thing that doesn't change.
Rest all is temporary.

In last 5 months I spoke to 50+ of you from LinkedIn. 

In those meetings, I realised there are many in the same place where I was 4 years ago.

It is the state of "self doubt", "constant confusion", "frustration". 

This state not only stops your career growth but also kills your personal life. 

Thats why I made it my mission to help software professionals in achieving their true career potential and  live fulfilling life.

I do it in 2 ways.

[1] By coaching on how to crack Big tech FAANG+ interviews (behavioural)  
[2] By coaching on how to get promoted (career growth).


I do it by sharing result and action oriented "systems" and "frameworks".  

These are proven frameworks that I have developed for myself & learned from others. 


Follow me (and press bell icon) for getting these systems and frameworks.

If you want advise or coaching for your unique situation, DM me "COACH"

Chandra Shekhar Joshi

If a Staff engineer is leading the technical project, why do you need an Engineering Manager?

Why not just fire the EM and save a $400k salary?

Many engineers think the EM is just there to update Jira tickets or approve vacations.

When a strong technical lead is in the room, the manager can feel redundant.

The real reason is the cost of indecision.

Companies do not keep EMs around for admin work.

The biggest risk in a large tech organization is not a high salary.

The biggest risk is the cost of indecision.

When smart engineers debate endless technical trade-offs, projects freeze.

This leads to massive execution failures.

This is where the EM earns their paycheck.

The EM's job is to drive decision making.

They must look at the project from the customer and business perspective.

To do this well, they must understand the technical nuances explained by the Staff engineer.

A strong EM bridges the gap between business needs and technical reality.

They do this by taking a few key actions:
• Listen to the technical options.
• Align those options with what the customer actually needs.
• Make the hard call to unblock the team.

When an EM refuses to make these decisions, the team struggles.

Timelines slip, and the project ultimately fails.

This is not about blocking technical innovation.

It is about preventing execution failure.

The Staff engineer figures out how to build it.

The EM makes sure it actually gets shipped.

Crack FAANG+ Sr., Staff+, EM Behavioural, and System Design HLD interviews | DM me “COACH” | Engineering Manager @ Amazon | Engineering Career Coach | FAANG+ Interview Coach

Treating a system design interview like a shopping list is the biggest reason candidates fail Senior and Staff+ roles.

"I’ll take one Kafka, a side of Redis, and some NoSQL because the data is large."

This "tool-first" approach is the fastest way to a rejection.

In top-tier companies, you aren't hired for the tools you know. You are hired for the complexity you can avoid.

Brilliant engineers get rejected because they force-fit "cool" tech into problems that a single-server monolith could solve better.

If you want to land a Staff-level role, you have to stop thinking in "boxes" and start thinking in "trade-offs."


I’ve put together a carousel breaking down the 4 most common architectural traps I see in interviews:
- Why Kafka isn't a "magic pill" for reliability.
- The truth about NoSQL vs. Scale.
- Why your "low latency" Redis plan might be a red flag.
- The danger of "resume-driven design."

Swipe through to see why these mistakes happen and how to fix them.

Save this for your next HLD round. 

And, if you are preparing for mid-senior/staff SDE or EM system design HLD interviews and need help, DM me COACH.

4 system design mistakes you should never make

Engineering leaders are getting AI agents wrong.

Everyone wants to use AI to write code fast.

But developers are afraid to ship agent-generated code to production.

They do not trust the output.

Without safety nets, shipping fast is risky.

The root cause is a focus on the wrong metric.

We are trying to make agents write perfect code.

But agents will always make mistakes.

The real value of an engineering team has shifted.

It is no longer about writing the code yourself.

It is about building strict guardrails for your non-negotiables.

Guardrails give your team the confidence to let AI write and ship code.

You must invest heavily in custom tooling and deep test suites.

Focus on two main areas.

Functional correctness:
- Protect the core business logic
- Build rigorous functional test suites
- Ensure the AI does not break critical workflows

Technical correctness:
- Robust security guardrails and key management  
- Define strict boundaries for system health
- Enforce strict system constraints

You cannot adopt agent-based development without this tooling.

Guardrails safeguard your business.

They allow your team to move fast and use AI safely.

Challenging the status quo is often just a lazy Staff Engineer's excuse to do "fun" work instead of "real" work.

The team has a 10-year-old monolith with 100+ complex features. It is messy. It is slow. The business needs a critical feature shipped in 3 months to hit a revenue target.

The Staff Engineer enters the chat: "We shouldn't just patch this. We need to challenge the status quo. Let's rewrite this into an event-driven, multi-region architecture."

Technically? They might be right. 
Practically? They are talking nonsense.

That plan requires 3 years of work. The business has 3 months.

The Staff Engineer feels great. They showed "Vision." 
They put "Led Architecture Modernization" on their promo doc.

Meanwhile, the team is thrown into chaos.
- Junior engineers waste weeks building POCs for tech, team will never use.
- Senior engineers get stuck in "Analysis Paralysis," writing docs for a future that won't happen.
- The actual feature, the one that pays the salaries, stalls.

This is not leadership, this is distraction.

If your proposal requires the company to pause reality for 3 years to build your dream architecture, you aren't being "strategic." You are being delusional.

True Staff-level thinking isn't about suggesting the Perfect solution. 
It is about suggesting the best solution within the business constraints.

Challenge the code, yes. But never ignore the clock.

The hardest "No" a new Staff Engineer has to say is to the manager who just fought to get them promoted.

You know the setup. 

Your manager went to bat for you. 
They built your promo case. 
You got the title. 

You feel incredibly indebted to them.

Two weeks later, they come to you with a timeline that is entirely unrealistic, or an architecture decision you know is fundamentally flawed.

This is where the trap snaps shut.

You know it’s a bad idea. But you say "Yes" anyway. Why? Because you don’t want to let them down. You don't want to look like you can't handle the new level.

When Staff Engineers operate out of this sense of obligation, the entire engineering org eventually pays the price. 

You build doomed solutions, tech debt piles up, and 12 months later, the whole team grinds to a halt.

If you read standard LinkedIn advice, people will tell you: "𝘑𝘶𝘴𝘵 𝘱𝘶𝘴𝘩 𝘣𝘢𝘤𝘬!" "𝘈 𝘨𝘰𝘰𝘥 𝘮𝘢𝘯𝘢𝘨𝘦𝘳 𝘤𝘳𝘦𝘢𝘵𝘦𝘴 𝘱𝘴𝘺𝘤𝘩𝘰𝘭𝘰𝘨𝘪𝘤𝘢𝘭 𝘴𝘢𝘧𝘦𝘵𝘺 𝘴𝘰 𝘺𝘰𝘶 𝘤𝘢𝘯 𝘴𝘢𝘺 𝘯𝘰!"

Let’s be real. That is textbook fantasy. Organizations are made of humans, and humans have egos. 

Your fear of damaging the relationship with the person who holds your career in their hands is 100% valid.

A blunt "No" triggers ego. But a blind "Yes" destroys the architecture.

𝗦𝗼, 𝘄𝗵𝗮𝘁 𝗶𝘀 𝘁𝗵𝗲 𝗳𝗶𝘅?

You cannot say a direct "No." And you cannot say "Yes." 

You have to stop sounding like an engineer who is complaining, and start speaking your manager's language.

That language is Risk.

Don't say: "𝘕𝘰, 𝘸𝘦 𝘤𝘢𝘯'𝘵 𝘥𝘰 𝘪𝘵 𝘵𝘩𝘪𝘴 𝘸𝘢𝘺, 𝘵𝘩𝘦 𝘢𝘳𝘤𝘩𝘪𝘵𝘦𝘤𝘵𝘶𝘳𝘦 𝘸𝘰𝘯'𝘵 𝘴𝘤𝘢𝘭𝘦."

Say: "𝘞𝘦 𝘤𝘢𝘯 𝘥𝘦𝘧𝘪𝘯𝘪𝘵𝘦𝘭𝘺 𝘣𝘶𝘪𝘭𝘥 𝘪𝘵 𝘵𝘩𝘪𝘴 𝘸𝘢𝘺 𝘵𝘰 𝘩𝘪𝘵 𝘺𝘰𝘶𝘳 𝘵𝘪𝘮𝘦𝘭𝘪𝘯𝘦. 𝘉𝘶𝘵 𝘐 𝘸𝘢𝘯𝘵 𝘵𝘰 𝘩𝘪𝘨𝘩𝘭𝘪𝘨𝘩𝘵 𝘵𝘩𝘦 𝘣𝘶𝘴𝘪𝘯𝘦𝘴𝘴 𝘳𝘪𝘴𝘬. 𝘛𝘩𝘦 𝘗𝘔 𝘮𝘦𝘯𝘵𝘪𝘰𝘯𝘦𝘥 𝘯𝘦𝘸 𝘳𝘦𝘲𝘶𝘪𝘳𝘦𝘮𝘦𝘯𝘵𝘴 𝘤𝘰𝘮𝘪𝘯𝘨 𝘪𝘯 6 𝘮𝘰𝘯𝘵𝘩𝘴. 𝘐𝘧 𝘸𝘦 𝘵𝘢𝘬𝘦 𝘵𝘩𝘪𝘴 𝘴𝘩𝘰𝘳𝘵𝘤𝘶𝘵 𝘯𝘰𝘸, 𝘸𝘦 𝘸𝘪𝘭𝘭 𝘩𝘢𝘷𝘦 𝘵𝘰 𝘳𝘦𝘸𝘳𝘪𝘵𝘦 𝘵𝘩𝘦 𝘦𝘯𝘵𝘪𝘳𝘦 𝘴𝘺𝘴𝘵𝘦𝘮 𝘵𝘩𝘦𝘯. 𝘐𝘵'𝘴 𝘨𝘰𝘪𝘯𝘨 𝘵𝘰 𝘣𝘦 𝘢 𝘮𝘢𝘴𝘴𝘪𝘷𝘦 𝘳𝘦𝘱𝘶𝘵𝘢𝘵𝘪𝘰𝘯𝘢𝘭 𝘳𝘪𝘴𝘬 𝘧𝘰𝘳 𝘰𝘶𝘳 𝘵𝘦𝘢𝘮 𝘸𝘩𝘦𝘯 𝘸𝘦 𝘩𝘢𝘷𝘦 𝘵𝘰 𝘧𝘳𝘦𝘦𝘻𝘦 𝘧𝘦𝘢𝘵𝘶𝘳𝘦 𝘸𝘰𝘳𝘬 𝘧𝘰𝘳 𝘢𝘯 𝘦𝘯𝘵𝘪𝘳𝘦 𝘲𝘶𝘢𝘳𝘵𝘦𝘳. 𝘏𝘰𝘸 𝘥𝘰 𝘺𝘰𝘶 𝘸𝘢𝘯𝘵 𝘵𝘰 𝘩𝘢𝘯𝘥𝘭𝘦 𝘵𝘩𝘢𝘵?"

𝗡𝗼𝘁𝗶𝗰𝗲 𝘁𝗵𝗲 𝘀𝗵𝗶𝗳𝘁? 

You aren't saying "I can't do it." You are pricing the risk.

Suddenly, it isn't You vs. Your Manager. 
It is You and Your Manager vs. The Business Risk.

If you want to survive as a Staff Engineer, you have to realize that you don't win arguments with technical correctness. You win them by making the wrong decision too expensive to choose.

Candidate: "I need to update two tables atomically. I chose Cassandra, so I'll just use a LOGGED BATCH."

Interviewer: "How does that handle isolation during the write?"

Candidate: "..."

This is a red flag for a senior role.

It tells, you are designing by keyword matching, not by understanding guarantees.

This is how it goes: 

1- You pick a database (like Cassandra) because "it scales." 
2- You realize halfway through the design that you need transactional logic. 
3- You google "Cassandra transactions" and find LOGGED BATCH. 
4- You plug it in to "make the architecture work."

This is dangerous engineering.
You are confusing Atomicity with Isolation.

In Cassandra, a batch guarantees that all writes eventually succeed (Atomicity). It does not guarantee that other clients won't see partial data while the batch is writing (Isolation).

If you build a financial system on this assumption, you will show users incorrect balances.

Good architecture should not depend on obscure features you don't fully understand. 

If you need ACID, use an RDBMS or a dedicated pattern like Sagas. Don't try to hack a NoSQL store into doing something it wasn't built for.

I broke down exactly how LOGGED BATCH fails at isolation in the carousel below.

Save it for later for your next system design interview.

If you are preparing for mid-senior/staff SDE, EM system design HLD interviews, and need help, DM me COACH.

Transactions in Cassandra

Most engineers hear "high read volume" in an interview and say, "I will add Redis." Jumping straight to caching shows a lack of depth. Here are 7 ways to solve hot reads before and after you add a cache:

# 1: Push the single node
• If possible, upgrade your physical hardware first.
• Build strategic indexes based on your exact access patterns.

# 2: Denormalize data
• Stop joining four tables for every read request.
• Duplicate data intentionally to create flat, read-optimized views.

# 3: Distribute the database load
• Route read traffic to your high-availability replicas.
• Shard the database to physically isolate intense read pressure.

# 4: Add smart caching
• Introduce in-memory storage like Redis.
• Use cache key versioning instead of simple time-to-live expirations to prevent stale data.

# 5: Mitigate extreme traffic spikes
• Use request coalescing so a sudden spike only hits the database once.
• Fan out hot cache keys by adding random suffixes to prevent network saturation.

# 6 : Invert the workflow
• Stop pulling and assembling heavy data on the fly.
• Pre-compute heavy reads asynchronously and push them to a cache on write.

# 7: Manufacture cache locality
• Group highly unique inputs like precise GPS coordinates or millisecond timestamps.
• Truncate them into shared buckets to force instant cache hits.
  
Knowing the definition of a cache is for novice.
Knowing the exact progression of read scaling is for staff engineers.

If you are preparing for mid-senior/staff SDE, EM system design HLD interviews, and need help, DM me COACH.

I scheduled a post on LinkedIn. The popup closed. A notification appeared at the bottom: "View scheduled post." I clicked it. My scheduled post didn't show in the scheduled posts list.

For a second, I panicked. Did I just lose my draft OR accidentally it got posted, instead of being scheduled?

I waited. Refreshed. 20 seconds later, it finally appeared.

I am not saying this is a critical "STOP-THE-WORLD" bug.

I am saying this is a classic disconnect between Engineering (Architecture) and Product (Functional Requirements).

I don't work at LinkedIn, but I have built enough systems to guess what changed.

Previously, this felt instantaneous.
Now, there is a clear lag.

My assumption is they moved the scheduling flow to be asynchronous.

They likely introduced a message queue (like Kafka) to handle the load.

The problem is, backend changed (Sync -> Async).
But the UI/UX didn't update based on the architecture. Or maybe it was assumed it would "just work."

- The UI still behaves as if the write is instant.
- It shows the "View" notification immediately.
- It sets an expectation that new backend cannot meet.

Why does this matter?

For a LinkedIn post, this is a minor annoyance.

But imagine this same pattern in a critical system.

You pay for an order.
The success screen says "Payment Complete."
You check your Order History.
It is empty.

This is not about lag. It is about user confusion and trust.

If this happens in a critical system, your user will get confused and never come back.
You are adding to their anxiety.

It is about keeping your Architecture and Functional Requirements in sync.

If you move to async, you must update the user contract:

𝟭- 𝗢𝗽𝘁𝗶𝗺𝗶𝘀𝘁𝗶𝗰 𝗨𝗫: Show the item in the list immediately, but mark it as "Processing..." so the user knows it exists.

𝟮- 𝗘𝘅𝗽𝗹𝗶𝗰𝗶𝘁 𝗺𝗲𝘀𝘀𝗮𝗴𝗶𝗻𝗴: If the list is empty, show a banner: "If you have scheduled your post just now, please check back in 1-2 minutes as it gets processed."

This is why I spend 60 minutes just teaching engineers how to think about requirements.

They are not a checklist you rush through.
They are the make-or-break constraints for your system design.

Engineers love adding Kafka.

But if you don't update the product behavior to match the architectural reality, you aren't designing a system.

You are confusing your users.

If you are preparing for mid-senior/staff SDE, EM system design HLD interviews, and need help, DM me COACH.

Linkedin's eventual consistency bug

I spent 100+ hours breaking down Databases... Here’s what I found.

I was obsessed.

I dove deep into 4 different databases - both SQL and NoSQL, to understand:

- How they’re created
- How they work internally
- What they’re actually made of

The deeper I went, the more complex it seemed.
But then I saw the pattern.

No matter how fancy or complex a database looks...
They all boil down to just 3 simple layers.
Yes, just 3 layers.

If you understand these 3 layers, you can:
- Understand any database
- Build a deep intuition about its architecture
- Answer any interview question confidently

Here are the 3 layers:

𝟭. 𝗨𝘀𝗲𝗿 𝗔𝗰𝗰𝗲𝘀𝘀 𝗟𝗮𝘆𝗲𝗿
This layer defines how users connect with the database.

It includes:
- SQL (Structured Query Language)
- HTTP APIs (For NoSQL databases)
- Specific Protocols (Database-specific query languages)

Examples:
- RDBMS: SQL for user access.
- Cassandra: CQL (Cassandra Query Language).
- NoSQL (DynamoDB, MongoDB): HTTP APIs for user access.

𝟮. 𝗥𝗼𝘂𝘁𝗶𝗻𝗴 𝗟𝗮𝘆𝗲𝗿
This layer routes read and write requests to the correct server or node.

This layer is optional for databases on a single node (e.g., most traditional RDBMS).

But for distributed databases (NoSQL), this layer is critical.

Examples:

- DynamoDB: Uses a request router that connects to partition metadata to route reads/writes to specific partitions.

- MongoDB: Uses mongos that connects to config servers for routing.

- Cassandra: Uses snitch to find the node where the data resides.

Pro Tip:

Almost all of them use the same routing algorithm, Consistent Hashing.

It efficiently distributes data across nodes for high availability and scalability.

𝟯. 𝗦𝘁𝗼𝗿𝗮𝗴𝗲 𝗟𝗮𝘆𝗲𝗿
This layer defines how data is stored on disk.

There are 2 common data structures used in this layer:

B/B+ Tree:
- Used in 80% of databases (both SQL and NoSQL).
- Provides consistent O(log n) read and write access.
- Perfect for range queries.
Examples: All SQL DBs, DynamoDB, MongoDB.

LSM Tree:
- Optimized for high write throughput.
- Uses concepts like SSTable, compaction, tombstone to manage data efficiently.
Examples: Cassandra, ScyllaDB.


That’s it.
All databases, SQL and NoSQL, are just a combination of these 3 layers.

What surprised me the most?

Despite all the marketing hype, SQL and NoSQL databases are fundamentally the same.

They just assemble these 3 layers differently.

Next time you see a new, fancy database...
Don’t get swayed by the buzzwords or social media hype.

Instead:
- Apply this 3-layer architecture.
- Break down the database into these 3 layers.
- Understand it in 2 minutes flat.

This mindset shift will make you immune to hype and unmatched in system design interviews.

If you need targeted help with cracking System Design HLD rounds in 4.5 hours, DM me COACH.

Most senior engineers talk about "consistency" as if it only has one meaning.

They use the same definition for a local SQL database and a distributed NoSQL cluster.

This is a red flag for interviewers.

The word is the same, but the context changes everything.
• In SQL, consistency is about ACID properties.
• In NoSQL, consistency is about read guarantees.
If you mix these up, you show a lack of distributed systems depth.

We are taught that consistency means "data is the same."

But in a single-node SQL DB, consistency means the data follows all schema rules and constraints.

It is about valid state transitions.

In a distributed NoSQL DB, you have multiple nodes across a network.

Consistency here is about how quickly a read reflects the latest write.

SQL Consistency (ACID):
• It ensures your database never enters an invalid state.
• Example: A bank transfer must debit one account and credit another.
• If one fails, the whole thing rolls back.

NoSQL Consistency (Distributed):
• It is a trade-off between latency and data freshness.
• Strong Consistency: You always get the latest data but suffer higher latency.
• Eventual Consistency: You get fast responses but might see stale data for a few milliseconds.

Next time an interviewer asks "How do you ensure consistency?", do not jump into an answer.

Ask this instead: "Are we discussing ACID compliance for data integrity, or read consistency across distributed replicas?"

That one question moves you from "coder" to "architect" in 10 seconds.

Stop memorizing definitions. 
Start understanding trade-offs.

If you are preparing for mid-senior/staff SDE, EM system design HLD interviews, and need help, DM me COACH.

Ever wonder why some senior engineers or managers are a nightmare to work with?

It is easy to assume they are just "bad" at their jobs or difficult people.

But usually, it is not a personality flaw.

It is a career scar.

When you spend your early years in a specific team culture, you don't just work in it.
You become it.

If your first manager was a micromanager, you learn that constant checking is how work gets done.

If your first senior engineer was dismissive, you learn that ego is part of being "elite."

You live in that environment for a decade, and it feels like the only way to survive.

The problem starts when you get promoted.

You move from "living in" an environment to "creating" one for others.

Most people don't stop to think about what kind of culture they want to build.

Instead, they simply impose the same environment they grew up in.

They pass down the same habits that once made them miserable as juniors.

Remember, scars are not lessons.

If something bad happened to you as a junior, and now you do the same to your team, that is not learning.

That is an unhealed scar.

A real leader looks at their past and decides which parts of their environment were actually helpful.

They choose what to keep and what to break.

If you find yourself being the person no one wants to work with, look back at your first five years.

You might be solving problems using a survival manual that is ten years out of date.

Growth is about unlearning the environment that raised you.

If you ask AI for a message queue, it will give you RabbitMQ code in 3 seconds. But... 

AI doesn't know the operational reality of your business. 

It won't warn you that at 2 AM, when you desperately need to replay failed messages, you will realize you picked a tool with no offset to reset.

This exact lack of foresight shows up constantly in system design interviews right now.

When the interviewer asks how to replay messages if a consumer fails, the candidate answers quickly.

They confidently state that in Kafka, we reset the offset, but in RabbitMQ, we cannot do that.

But when the interviewer asks why, the candidate goes completely silent.

This is a red flag from a Staff candidate.

They memorized the API, but they completely missed the core storage model.

Knowing the features of a tool is no longer a competitive advantage.

Here is the staff-level breakdown of what is actually happening under the hood.
 
Kafka uses a "dumb log" model.
 
It is a passive, append-only file on disk, implemented as partitioned segment files with retention and optional compaction policies.
 
Think of it like a massive array.
 
The consumer holds an index pointer called an offset, which is stored and managed via the __consumer_offsets topic.
 
If you want to retry or replay data, you simply move the pointer back.
 
The log itself does not care.
 
RabbitMQ uses a "smart queue" model.
 
It is not a passive file.
 
Every single queue is managed by its own active Erlang process.
 
This process acts as a dynamic, in-memory state machine, while persisting messages to disk if the queue and messages are configured as durable.
 
When a consumer is free, the queue process dispatches a message.
 
It changes the message state from "ready" to "unacknowledged" and starts a timer.
 
When the consumer finishes the work, it sends an ACK.
 
The queue process receives the ACK and permanently deletes the message from its list, meaning replay is not possible after ACK unless the message was routed to a dead-letter exchange or stored externally.
 
If the consumer crashes, the timer expires.
 
The process simply moves the message back to the "ready" state for the next worker.
 
There is no array.
 
There is no offset pointer to reset, because RabbitMQ uses an ACK/NACK-based delivery model rather than offset-based consumption.
 
Once an ACK is received, that message is gone forever, unless alternative patterns like DLQ reprocessing or RabbitMQ Streams are used.
 
Anyone (including AI) can tell what a tool can do. But Staff engineers know how it is built, and when to use it.
 
If you are preparing for mid-senior/staff SDE, EM system design HLD interviews, and need help, DM me COACH.

Designing for the "future" is a common trap in system design interviews.

Engineers often think they are being smart by adding flexibility.

But they end up breaking the actual system they were asked to build.

I see this failure in almost every senior-level interview.

Most candidates design for:
• Imaginary scale that isn't in the requirements.
• Hypothetical users that don't exist yet.
• Future features that the interviewer hasn't asked for.

They say things like:
• "Let's use Kafka here so other systems can consume this data later."
• "Let's make this call async to support more use cases in the future."
• "Let's build a separate service because it will be more reusable."

When you ask why, the answer is always: "This will support more things later."

Here is what actually happens when you prioritize the future over the present:
• You add Kafka when a direct database write was all that was needed.
• You force an async flow when the product required a sync update for a better user experience.
• You increase latency just to look "scalable."

You ended up solving ten future problems but failed to solve the one P0 problem on the whiteboard.

The problem is that many engineers value extensibility over correctness.

They add async queues for "eventual consistency" when the core requirement was strict ordering or immediate confirmation.

They build for edge cases that might never happen.

Extensibility that breaks the current requirement is not a good strategy. It is a distraction.

How to design like a staff engineer
If you are in an interview, or even designing at work, follow these rules:
• Nail the core use case (P0) before anything else.
• Stick to the non-functional requirements (NFRs) you agreed on.
• Only optimize for the future if that future is explicitly in scope.

If your "future-proof" design makes the current system slow or inconsistent, it is a bad design.

Interviews are about solving the problem in front of you, not designing a fantasy.

If you are preparing for mid-senior/staff SDE, EM system design HLD interviews, and need help, DM me COACH.

You’ve probably innovated this year, and still can’t answer the behavioural interview question : “Tell me about a time you innovated.”

Most engineers don't know what to answer when this question comes up in interviews. 

Not because they haven’t done anything innovative, but because they don’t know how to talk about it.

The moment the interviewer asks: 
“Why do you think that was innovative?” …it gets difficult:
- Blank stares
- Fumbled answers
- Vague justifications

Disconnect is, everyone at work says they “value innovation.” 

But no one stops to label your actual work as innovative.

You:
- Built a workaround that unblocked your team in a day
- Automated a process that saved 3 hours per deploy
- Reimagined a user flow that cut churn by 20%

But since it wasn’t flashy or groundbreaking… you hesitate.
Instead, you assume innovation = “build ChatGPT” or “launch a rocket.”

“Is that automation really innovation?” Yes. 
If it drove impact in a novel way, it counts.

Innovation isn’t about buzzwords or billion-dollar launches. 

It’s about: 
- Solving the right problem 
- In a clever, unexpected, or high-leverage way 
- That creates real-world results

How to nail this question?

[1] Start with the problem. 
What was painful, slow, or broken? 

[2] Show your thinking. 
What made your approach different? 

[3] Quantify the outcome. 
What improved, and by how much? 

[4] Frame the innovation. 
Why wasn’t this just business as usual?

Example: “Sign-ups were dropping 15% due to slow KYC. Backend fix would take 3 months. I proposed a lightweight pre-verification to catch errors earlier. Took a week to build. Reduced verification failures by 60%, conversions rebounded. It was a simple, innovative solution that hit the root cause fast.”

Innovation isn’t about what you built. 
It’s about how you thought differently, and what that unlocked.

Next time they ask, “Tell me about a time you innovated”:
- Don’t freeze
- Don’t fumble
- Own it. Frame it. Prove the impact.

If you are preparing for mid-senior/staff SDE, EM behavioural interviews, and need help, DM me COACH.

99% of the engineers who cracked behavioural interview round, have these 10 similarities.

They:
- All of them used answering frameworks like SBI, STAR.
- Explained "Situation" in SBI from business perspective.
- Explained Situation, Impact in less than 1 minute each.
- Added time window to make answer more trustworthy.
- Answered  their career stories without any acronyms.
- Described failure quickly and spent time in learnings.
- Were ready with impact numbers for all their stories.
- Explained what did they try to recover from failures.
- Explained how they collaborated with the team.
- Used  “I” , instead of the commonly used "we".

Small nuances in interviews,
But huge differences in outcomes
Read one more time, to crack your next one...

anything else would you add to the above list?

If you are preparing for mid-senior/staff SDE, EM behavioural interviews, and need help, DM me COACH.

75% of employers use behavioral interviews to predict your future success.

But most senior engineers fail because they overcomplicate their answers.

Save this post for your next behavioral interview.

Check the attached whiteboard breakdown to see exactly how the 4-step process works.

Here is how to apply those rules in a real interview setting.

The why
• My e-commerce team saw a sudden drop in checkout conversions.
• Rates fell from 65% to 48% in one week.
• We found the payment page load time had jumped from 1 second to 4 seconds.
• Users were frustrated and abandoning their carts.
  
The what and how
• I checked our frontend metrics using New Relic and DevTools.
• I found a new third-party analytics library blocking the page render.
• I proposed a fix to defer the script using lazy loading.
• I collaborated with QA and Legal to ensure we stayed compliant.
• I rolled the fix out slowly using a feature flag to monitor errors.
  
The impact
• Page load times dropped back down to 1.2 seconds.
• Our conversion rate recovered to 66% within three days.
  
Notice the pattern?
Zero complex frameworks.

Just a clear business problem, my specific actions, and hard numbers.

If you are preparing for mid-senior/staff SDE, EM behavioural interviews, and need help, DM me COACH.

I keep seeing a strange new line in engineering job descriptions. It says, "If you only want to write code, this job is not a fit."

This is a ridiculous statement.

It is like a plumbing company posting an ad that says, "If you only fit pipes, this is not for you. We need water to flow."

Who ever said software engineering was just about typing code?

Code was always a commodity.

It just used to take a lot more time to produce that commodity.

Now, that time equation has changed.

But the core of the job has not changed at all.

Software engineering is about thinking.

It is about making decisions.

Most importantly, it is about delivering outcomes.

That is how it was 30 years ago.

That is exactly how it is today.

Many engineers merge their pull request and move on. 
They assume QA team will catch the bugs.

This mindset keeps you stuck at the mid-level. 
The real work starts after the code is deployed in production.

If you do not own the outcome, you are just typing syntax. 

Even with AI writing half the code today, you still hold the pager.

Senior engineers operate differently. 

They take full accountability for the system from start to finish.
• They monitor the metrics immediately after a launch
• They anticipate how a failure impacts the end user
• They fix the broken process instead of just patching the bug

Ownership is not an assigned role. It is a daily choice.
Take responsibility for the whole system, and your career will grow.

Most people at work are waiting for an excuse to stop. They look for a valid reason to fail, and not do it.

And in a large engineering organization, finding a reason is easy.
- The requirements were too vague.
- The dependency team was slow.
- The legacy code was a mess.

These are all valid. They are logical. And they are the exact reasons why most people and product don't reach anywhere. Yes, they work, but they "only work".

They use these reasons to justify why something did not happen. It protects their ego. But it kills ownership.

Imagine this is the last piece of code you will ever write. 
The last system you will ever design. 
The last decision you will make before you die.

Would you still accept that delay? 
Would you still point fingers at the messy legacy code? 
Would you still wait for permission?
Would you still look for reasons for, why will it fail?

Or would you find a way to make it work, no matter what?

The "valid reasons" will always be there, as long as you believe in - "you will live tomorrow".

Don't try understanding the literal meaning of these words. Try understanding what can't be said in any language, directly.

How to get data model right in system design?

You can change your tech stack later. 
But if your data model is wrong, your system is set up to fail.

Many engineers spend their early years focused only on business logic. 
They rarely design data models for complex systems from scratch.

Then they step into a system design interview and freeze. 

Their proposed models require complex in-memory joins to make the system work (anti-pattern).
 
This quickly exposes a lack of real experience.

A common mistake is thinking data modeling is just about scale. It is not. 

Many scaling problems have nothing to do with data models at all.

Another mistake is obsessing over saving storage space. Or engineers try to build a perfectly normalized view. This points you in the wrong direction.

How to get it right? 

To build a great data model, you must focus on access patterns. 

Work backward from how the data will be used by the application. 
Do not focus on saving disk space.

How to practice? 

You do not need massive scale to practice data modeling. 

If you lack complex data challenges at work, practice with these domains:
- Trip booking platforms
- Hotel reservation systems
- E-commerce flows

Think through the entire system end to end. 
Map out the data for inventory, checkout, payments, and orders.

Design for how data is read and written. 
The right access patterns will always lead to the right data model.

If you are preparing for mid-senior/staff SDE, EM system design HLD interviews, and need help, DM me COACH.

taplio

The Battle

Lucie Lachaux

Chandra Shekhar Joshi

Followers

Posts

Engagements

Likes

Performance ScoreLucie Lachaux's LinkedIn Performance

Performance ScoreChandra Shekhar Joshi's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Lucie Lachaux

Chandra Shekhar Joshi

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Lucie Lachaux's LinkedIn Performance

Performance ScoreChandra Shekhar Joshi's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreLucie Lachaux's LinkedIn Performance