Amazon

As an engineering manager in Amazon AppStore, I work on defining and executing the strategy and roadmap of my business. I lead a team of 10+ amazing engineers who work backwards from customer problems. We build scalable solutions using cutting-edge tools from AWS services, Machine learning, and Generative AI. Our products have reduced the time taken in application testing by 5 times using machine learning-based solutions. 
As a manager, I have coached 100+ engineers in their career development and built a strong team of problem-solving, self-driven individuals.

Software Development Manager

Leading the engineering team, that powers the backend of the Product Details page of Amazon shopping site worldwide, top 10 visited page on internet.

Goldman Sachs

Leadership | Hiring & Mentoring | Big Data | Data pipeline

Designed and managed a team for strategic rebuilt of conflict management system to support the new advanced nature of business with futuristic tech stack using Elasticsearch, microservices & Angular etc. It uses event sourcing in UI to create a reactive front end using RxJS.

Architected & managed a team to optimize the rule based tagging to highlight potential risks related to deals using event driven architecture. This improved the existing rule based system and reduced the noise by 80%.

Managed a team to create a self-service reporting dashboard using Elasticsearch & Angular.  This product saved 208 business days/year for business.

Executed a learning and training program for Legal, Compliance and Audit business unit by Chairing learning and training committee. This committee delivered 500+ person hours of training and mentorship

Vice President

Web & Mobile | Event Sourcing | Mentoring | Training 

Managed a team of interns to create a product for automating various manual approval processes by leveraging JBPM & microservices using RESTful architecture. This system currently hosts 100+ business processes and over 1M unique instances of those processes.

Influenced the business to get funding for a project by implementing a system showcasing how emails can be connected with the application to track the conversation which led to a decision for a deal.

Applied Event sourcing with RabbitMQ and ElasticSearch in implementing a data pipeline for conflict management system so that users can execute searches & reports in near real time.

Led & designed a workflow to document the decision taken during conflict approval process. Further improved the workflow by extending it to junior users, without compromising the risk associated with approval. This led to 70% faster issue approvals.

Associate Programmer

Analyst programmer

Accenture

Web | Enterprise Application Development | Agile | TDD 

Reduced the coding time by 20% by creating a reusable UI component using Google Web Toolkit, leading to a 70% reduction in coding time in similar requirements. 

Redesigned the architecture of a GWT based page, which lead to 70% less defects.

Automated the copy/paste of Review comments from word file to an excel file using POI library, leading to a 100% reduction in developer’s time.

Performed the role of Subject Matter Expert for GWT and Selenium testing tool in Accenture.

Created a team to organize technical sessions and discussions for the project, leading to a decline of 30% questions/doubts from the new joiners and other team members.

Senior Programmer

Favworks

Application Developer | Customer Relationship Management Tool | Billing Tools 

Reduced the time taken in organizing the company’s billing/client/project information by 80% by developing a Billing Software, which was later launched as one of the company’s product.

Developed a CRM and Social Network from scratch using PHP, MySql and JQuery.

Software Engineer

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"

You aren't paid a Staff Engineer salary to write code faster than a Senior.

If that was the goal, they would just hire two Seniors.

Or buy an AI tool.

Yet, I often see critical teams of 15 people with 3 or 4 Staff Engineers.

On paper, this ratio makes no sense. In reality, it is the only thing keeping the business alive.

When you own a critical business function, things break. High-traffic systems go down. Deadlines get tight.

The team starts panicking.

You are not there to close tickets during the calm times. You are there to make the right call during the chaos.

You ensure the team doesn't make a suboptimal decision that feels good today but kills the architecture next year.

You exist to stop the technical debt from becoming a bottleneck that halts business growth.

But still, this is where many of you Staff engineers get it wrong.

Many of you join a new team or get promoted as a Staff Engineer and immediately miss this point.

Instead of stopping the bleeding, you start your first month by:
- Complaining about the existing legacy system.
- Asking why we aren't doing it the way "Company X" did it.
- Philosophizing about how AI will replace the whole team soon.
- Proposing a total re-architecture without understanding the business context.

This is just noise.

There are already enough people in the company adding noise. They didn't hire a Staff/Principal Engineer to add more of it.

Your value is not in the code you write. It is in the clarity you bring.

You are there to ensure the team, including the Engineering Manager and Product Manager, does not get lost in the noise.

You guide them to make decisions that allow the business to scale.

You are not paid for the features you ship. You are paid for the catastrophic mistakes you prevent the team from making.

Senior Engineers build the product. Staff Engineers protect the business.

Don't confuse the two.

Chandra Shekhar Joshi

Crack mid-senior/staff level FAANG+ SDE and EM Behavioural, and System Design HLD interviews | DM me “COACH” | Engineering Manager @ Amazon | Engineering Career Coach | FAANG+ Interview Coach

Interviewer: "The system must have low latency."
Candidate: "Got it. I'll use Redis for caching."

This is a weak answer, especially for a Senior/Staff engineer. Here’s why.

No interviewer has ever asked you to design a high latency system. "Low latency" is a given. It is not a useful requirement.

The real question is: How low is low?

Is the p99 target 500ms? 100ms? 20ms? 

You cannot design a solution until you quantify the problem.

Jumping straight to "Redis" is a, keyword-matching reflex.

The Senior+ level approach is to first understand the actual target. 
Then, you design a baseline system.

If that baseline doesn't meet the latency target, you optimize.

And caching is almost never the first optimization.

You check these first:
- Can I improve the database schema to remove expensive joins?
- Can I add read replicas (leader-follower) to scale the reads?
- Can I use a different database (like a KV store) that guarantees single-digit-ms latency?

Only when those options are not enough, you introduce the complexity of a cache.

When you hear the key "low latency" and just return the value "Redis"... 
You're not a system designer.
You're a human key-value store.

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

Interviewer: "Design a live stream for 10 million users."
Candidate: "I'll use Kafka. The user devices will be the consumers."

This is an immediate fail. It shows you don't understand how Kafka actually works.

When asked about the setup, the candidate often says: "The infra team owns the Kafka cluster."

That is a lazy answer. Infra owns the servers. You own the topology (partitions, keys, consumer groups).

If you don't define the partitions and consumer groups, you build a broken system. 

Here is the math:

Scenario 1: One Consumer Group. 
You have 1,000 partitions and 10 million users. Kafka logic dictates that extra consumers sit idle. Result: 1,000 users watch the match. 9,999,000 users see a black screen.

Scenario 2: 10 Million Consumer Groups. 
You give every user their own group. Now everyone can read the data. Result: You create a metadata storm that crashes the cluster.

And logically? You are connecting 10 million public devices directly to your internal backend. 

That isn't a design. It is a self-inflicted DDoS attack.

Live streaming relies on CDNs (Edge servers), not backend logs.

When you hear "Live Stream" and blindly produce "Kafka"... 
You aren't going to consume an offer letter.
Instead, you are sending your offer letter straight to the Dead Letter Queue.

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

Engineer: "I tried using AI for the new shipping feature. It wrote great unit tests, but the core logic is broken. It hallucinated APIs that don't exist."

Leader: "Did you prompt it correctly?"

Engineer: "Yes. I uploaded the architecture docs and the context files."

Leader: "You need to be more prescriptive. Give it specific examples of the pattern we use."

After 2 days...

Engineer: "I spent ten hours doing that. It still tried to use the wrong framework. Added code and features that were not required."

Leader: "Well, did you provide the Low-Level Design with code snippets?"

Stop and look at that last request.

If I have to write the code to tell the AI how to write the code, I might as well copy-paste it into my IDE.

This conversation is happening in many engineering orgs right now.

It is an example of the sunk cost fallacy.

Leaders are invested in the narrative that AI will make organizations "100x more productive."

They cite examples like, "My 2-year-old built Tic Tac Toe in 2 minutes."

But there is a massive difference between generating a static game and building a production-ready feature in a complex distributed system.

When the tool fails, the narrative forces you to blame the instruction, not the tool.

So you make the task less ambiguous. You break it down. You feed it context.

You end up treating the AI like a Staff Engineer who needs tasks broken down like an intern.

If you hired a Staff Engineer who needed that level of hand-holding, you would fire them.

You wouldn't say, "I just need to prompt them better."

Use the tool for what it is actually good at.

It crushes unit tests. It writes great boilerplate. It explains regex perfectly.

Keep the narrative there.

But stop pretending it can architect complex systems with minimal to no hand-holding.

Real productivity comes from using the right tool for the job, not forcing the job to fit the tool.

Interviewer: "We save to the DB and publish to Kafka. What if one fails?" 
Candidate: "I will use a Two-Phase Commit (2PC) to ensure they both succeed."

This is a red flag and it shows the lack of understanding.

You just proved you treat infrastructure like magic.

Two-Phase Commit is not a config switch. It requires a Coordinator node. It requires all participants (DB, Queue, Search Engine) to implement a specific locking protocol:
- Phase 1: "Are you ready to commit?"
- Phase 2: "Commit now."

Yes, distributed databases like MongoDB support transactions using 2PC. 

But that works because the database vendors wrote the complex 2PC logic internally to coordinate between their own nodes. They implemented the protocol for you.

But your custom architecture? Your Elasticsearch cluster? Your Kafka topic? They do not natively support an external coordinator locking their resources.

To make 2PC work here, you have to implement the locking protocol wrapper for every single service. You are signing up to build a distributed transaction manager from scratch.

Even if you build it, you create a massive risk. 2PC is a blocking protocol. If the Coordinator fails, your resources stay locked. Your latency spikes. Your availability drops. You need to handle all of this. 

A Staff Engineer solves this by changing the constraints.

1. They aim for reliable eventual consistency: 
Does the user really need the search index to update the exact millisecond the record is saved? Rarely. A few seconds of lag is acceptable.

2. They use asynchronous patterns (Outbox or CDC): 
Don't write to two systems synchronously. Commit your data to the primary database first. 

Then, use the Transactional Outbox Pattern or Change Data Capture (CDC) to stream that change to Elasticsearch asynchronously.

3. They use Sagas break the flow into local transactions: 
If step 2 fails, trigger a "Compensating Transaction" to undo step 1.

Stop reaching for the most complex tool in the box.
Stop trying to fight the CAP theorem with complex protocols.

You aren't building a distributed system. You are building a distributed deadlock.

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

Interviewer: "Tell me about a time you handled a conflict."
Candidate: "Sure. So, we had this microservice, and it was using gRPC, and my manager wanted to use REST, but the other team was using Kafka..."

...three minutes later, you're still explaining the tech stack.

The interviewer is nodding, but they've checked out.

This is an immediate red flag.

Your answer isn't an answer. It's a "brain dump."

You are "thinking on the fly" and beating around the bush.

When a candidate uses endless, random technical examples to explain a situation, it signals two major failures:
- You don't understand the core situation well enough to be concise.
- You have a communication problem and are using examples as a crutch.

You're hiding in the tech details because you haven't identified the actual business or customer problem.

By the time you finally get to the point (if you ever do), the interviewer has already lost interest and stopped collecting useful data.

A senior-level winning answer is precise. It gets to the point fast.

It leads with the business context, not the tech stack.

Weak: "We had a problem with our Kafka consumers lagging..."

Strong: "My manager and I disagreed on a launch deadline. The conflict was over high customer risk vs. a hard revenue deadline."

See the difference?

The situation and its impact come first. The technology is just supporting context, not the main story.

Structure your answer. Deliver the core point in ~3 minutes or less.

Then, stop talking. Let the interviewer ask for more details if they want them.

If you are giving 10 minute monologue before coming to the point, it is not a good sign of seniority and communication. 

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

"Software engineering is done." Yet they just acquired Bun.

If the AI can do it all, what are they buying?

Here is the timeline of hypocrisy from Anthropic leadership:
- March 2025: Their CEO suggests AI will write 90% of code within 3-6 months, and 100% in 12 months.
- November 25: Another leader claims software engineering is "done" by the first half of next year (check screenshot).

The narrative is clear: Engineers are obsolete.

Then, on December 3, they announce the acquisition of Bun.

For context, Bun is a JavaScript runtime. It is complex, high-performance code built by talented software engineers.

This move destroys their own argument.

If Claude Code is so amazing that it replaces every developer, why buy Bun?

Why not just ask the AI to write a better runtime? If the model is truly that capable, generating a JavaScript runtime should be free.

You might argue, "They bought the team, not the code."

That makes it worse.

If software engineering is dead, why are you hiring software engineers?

Why pay a premium for a team that builds developer tools if the profession is ending in six months?

It does not matter if they bought it for the code or the talent. In either case, their checkbook proves their public statements are garbage.

There is zero hand-eye coordination here.

Stop listening to the hype. Watch where they spend their money.

Software engineering is not going anywhere.
Roll up your sleeves, and get to work.

Interviewer: "Tell me about a time you worked with a difficult stakeholder."
Candidate: "The Product Manager changed the requirements halfway through the sprint. I told them it was too late, proved the original plan was better, and we shipped as per the original plan."

This is a red flag for a senior role.

You told a story about "winning."

If you won, that means your stakeholder lost. That is not collaboration. That is an argument.

You also failed to define what "difficult" actually means.

A stakeholder changing requirements once is not being difficult. That is just the nature of business. Priorities shift. New data comes in.

If you think a single change request makes someone "difficult," you might be the difficult one.

Here is the winning Staff-level approach.

To answer this correctly, you must first establish a pattern.

Difficulty is not a one-time event. It is recurring behavior.
- A PM who changes scope two days before launch, five sprints in a row.
- A Staff Engineer who re-opens the same design debate for the 10th time without adding new data.

That is a difficult pattern.

Once you establish the pattern, do not talk about how you shut them down.

Talk about:
- How you identified the root cause of their behavior (empathy).
- How you set up a process to mitigate the disruption.
- How you secured a meaningful outcome, even if it wasn't perfect.

The interviewer is not checking if you can debate.

They are checking if you can work with humans who are stressed, disorganized, or stubborn without becoming toxic yourself.

Juniors try to win the argument. Staff engineers try to save the relationship and the delivery.

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

Interviewer: "Tell me about a time you traded quality for speed."
Candidate: "We skipped unit tests for three files to meet a deadline, but I added them right after launch."

This is a weak answer from a senior candidate.

It tells interviewer you are playing it safe.

This is because, you are afraid that admitting to "low quality" makes you look like a bad engineer.

So you pick a trivial example, skipping a few tests, that has zero real risk.

But this backfires.

It signals that you haven't been in a position where you had to make a real, painful trade-off.

It suggests you haven't carried the weight of a business-critical launch where "perfect" wasn't an option.

The 'Staff-Level' solution:

Stop trying to prove you are a "clean coder." 
Start proving you are a business partner.

A Staff engineer doesn't talk about skipping tests. 
They talk about skipping architecture.

A strong answer looks like this:
- We had a hard deadline for the Black Friday sale.
- The "right" design was a microservice with a dedicated DB.
- We built a monolith extension on the legacy DB instead.
- We accrued technical debt.
- But we captured $5M in revenue that would have been lost.

Why is this works?

It shows you understand that shipping value > perfect code.

It shows you can calculate the cost of debt vs. the cost of delay.

It shows you have the confidence to make a mess intentionally to win.

Junior engineers fear technical debt. Staff engineers manage it as a financial and business decision.

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

Interviewer: "We are designing a distributed KV store. Imagine a client has a 'Product' table with very low write traffic. Yet, they are seeing write timeouts."

Candidate: "But... how? If the write throughput is low, the system should easily handle it. Is the network down?"

The interviewer just marked you as "Junior."

You are designing logical isolation. 
You should be designing physical reality.

You are assuming that because the 'Product' table is logically separate, it lives in a vacuum.

In a distributed system (like the one you are designing), a single physical server (Node) hosts hundreds of partitions from different tables.

This is what is happening:
1. Node A hosts a partition for the 'Product' table (1 write/sec).
2. Node A also hosts a partition for a high-velocity 'Order' table (100k reads/sec).

Even though the 'Product' table is quiet, Node A is screaming.
- The disk I/O is saturated fetching Analytics data.
- The CPU thread pool is exhausted.
- The NIC (Network Interface Card) is at capacity.

The tiny 'Product' write arrives at Node A, finds no resources, and times out.

As the Sr/Staff engineer, you must design for tenant isolation.

When a node detects saturation, it must identify the "Noisy Neighbor" (the Orders partition) and migrate it to a different, underutilized node.

Your client doesn't have a code problem.
They have a bad roommate (Noisy Neighbor).

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

Memorizing "Design Uber or Google Drive" is the fastest way to get rejected for a Staff role.

If you have 15 days left, stop reading solution guides.

That strategy works for juniors. It fails for senior roles.

When you are short on time, you must shift from memorizing features to mastering trade-offs.

Focus on these 6 pillars instead.

𝟭. 𝗥𝗲𝗾𝘂𝗶𝗿𝗲𝗺𝗲𝗻𝘁𝘀 𝗮𝗿𝗲 𝘁𝗵𝗲 𝗳𝗼𝘂𝗻𝗱𝗮𝘁𝗶𝗼𝗻 Do not rush this. Get the functional requirements clear first. Then, map them to non-functional requirements. Crucial point: Do not use textbook definitions. Explain what "consistency" means specifically for a chat app versus a payment system.

𝟮. 𝗔𝗣𝗜 𝗱𝗲𝘀𝗶𝗴𝗻 𝘀𝗵𝗼𝘄𝘀 𝗺𝗮𝘁𝘂𝗿𝗶𝘁𝘆 Interviewers judge this heavily. It is not about the tools. It is about craftsmanship. Focus on resource naming, correct HTTP verbs, versioning strategies, and cursor-based pagination.

𝟯. 𝗗𝗮𝘁𝗮𝗯𝗮𝘀𝗲𝘀: 𝗚𝘂𝗮𝗿𝗮𝗻𝘁𝗲𝗲𝘀 𝗼𝘃𝗲𝗿 𝗰𝗵𝗲𝗰𝗸𝗹𝗶𝘀𝘁𝘀 Stop making lists like "Uber uses Postgres." Instead, build knowledge around guarantees. Does the database offer strong consistency? How does it handle write throughput? Map the database capabilities to your specific access patterns.

𝟰. 𝗠𝗲𝘀𝘀𝗮𝗴𝗶𝗻𝗴 𝗶𝘀 𝗮𝗯𝗼𝘂𝘁 𝘁𝗿𝗮𝗱𝗲-𝗼𝗳𝗳𝘀 Do not just pick Kafka because it is popular. Understand the difference between a distributed log and a message queue. Choose based on your specific need for strict ordering, HOL blocking, or raw throughput.

𝟱. 𝗧𝗵𝗲 𝗱𝗲𝗲𝗽 𝗱𝗶𝘃𝗲𝘀 You will get asked about the hard parts. Prepare for distributed transactions, consensus, and locking. Do not just define them. Explain how using a specific pattern solves a race condition in your design.

𝟲. 𝗔𝗱𝘃𝗮𝗻𝗰𝗲𝗱 𝗽𝗮𝘁𝘁𝗲𝗿𝗻𝘀 To clear the bar for Staff, you need to handle the edge cases. Study specific patterns for handling hot partitions, write-heavy workloads, and advanced data modeling.

Memorizing questions gets you through the screening. 
Mastering principles gets you the offer.

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

A few months ago: "AI will replace all software engineering jobs."
Now: "AI is a useless bubble that just hallucinates."

Both narratives are extreme. Both are wrong.

This is the real problem. We operate in extremes.

We want to label everything as 100% "good" or 100% "bad." We avoid the complex, nuanced middle because we crave quick labels.

The "replace everyone" story was a funding-driven hype cycle. Building useless agents for "hello world" tasks was noise.

But the new narrative, "AI is useless," is just as lazy.

I said it then and I'll say it now: AI is changing the way we work in software.

It’s not about replacing engineers. It’s about changing the work itself.

Tedious work that used to take days now takes a few hours. That is the real impact.

Stop thinking in extremes. The truth is not "good" or "bad." The truth is in how the tool is applied.

One out of five engineers treats job titles like a candy shop.

They think they can walk in and just "pick" the Staff Engineer or Engineering Manager role.

This is living in delusion.

If you are operating as an SDE 2, you cannot just apply for a Staff role and hope to crack it.

If you are a Senior Engineer, you cannot simply "decide" to become an Engineering Manager in a new company without prior experience.

The market has shifted.

During the hiring craze of 2021, companies were desperate. They took bets on potential. 

They hired Tech Leads directly into EM roles hoping they would figure it out.

That era is over.

Today, supply is high and open roles are scarce. No company will hire you as a Manager if you have never managed a team.

And even if you somehow crack the interview, the danger is real.

On day one, the company will expect you to perform at that level. If you have never handled that scope or ambiguity before, your chances of success are slim to none.

Titles exist for a reason.

They are not just labels. 

They define:
- The scope of problems you solve.
- The level of ambiguity you handle.
- The impact you deliver.

You cannot fake this scope in an interview.

Here is the reality check.
If you want to jump from IC to Manager, or SDE 2 to Staff, do not apply outside yet.

Do it in your current company first.

Work with your manager to expand your scope. Take on the responsibilities of the next level. Get the internal transition or promotion.

Build the proof of work first.

Roles are not candies distributed at the entrance. They are recognition for the work you have already done.

Interviewer: "Tell me about the most complex project you've led."
Candidate: "Sure. So on Project Avenger, we migrated the Hydra component to the new cluster because of high latency..."

This is a red flag for Staff+ roles.

By the time you finish that sentence, the interviewer has already checked out.

The root cause is, you are confusing "complexity" with "obscurity."

You assume the interviewer cares about your internal project names. They don't.

Conducting an interview isn't like watching a Netflix movie. It isn't inherently entertaining.

If you dive straight into the technical details without setting the stage, you turn a strategic discussion into noise.

You are treating the interview like a stand-up update, rather than a pitch to a stranger.

This round is a hybrid. It is 30% Behavioral (business impact) and 70% System Design (technical execution). You cannot fail either part.

Here is the Staff+ solution.

Stop dumping data. Start selling the story.

1. Stop using code names:
Your internal jargon means nothing outside your company. Translate everything immediately.

- "Project Avenger" becomes "The Global Payment Migration."
- "The Hydra Component" becomes "The Legacy Payment Gateway."


2. Tell why, whatever you are solving is important (The first 5 mins):
Before you draw a single box, tell them the stakes.

- What was the customer pain?
- How much money were you losing?
- Why did this project have to happen?
- Example: "We were losing 5% of checkout traffic due to latency. This project recovered $10M in annual revenue."

3. Now explain architecture (The next 25 mins): 
Now, treat it like a System Design interview.

- Explain the constraints you faced.
- Walk through the high-level design.
- Defend your trade-offs.

Junior engineers explain what they built. Staff engineers explain why it mattered.

If you are preparing for mid-senior/staff SDE, EM interviews, and need help with most recent complex project discussion, DM me COACH.

System design is not about memorizing every tool in the ecosystem.

It is about knowing how to use a few core concepts to solve massive problems.

Most engineers memorize "Kafka" or "Redis" without understanding why they need them.

This is a junior mindset.

Tools like "Kafka", "Redis", or "Cassandra" are just implementation details. They change every few years.

The physics of data movement do not.

If you want to pass the Staff bar, stop memorizing feature lists. Focus on these 5 invariants:

1. If you look for one bottleneck, look at the Database. 
-> Stateless services are easy to scale. You just add more servers.
-> State is hard. The database is almost always the constraint. If you aren't optimizing the state, you aren't solving the problem.

2. If you master one scaling pattern, make it Partitioning + Replication. 
-> Caching solves latency. It does not fix write throughput.
-> Partitioning (sharding) combined with Replication is the only way to scale write-heavy systems indefinitely while keeping data durable.

3. If you do one calculation, calculate Data Storage. 
-> QPS is easy to scale. Storage is a hard constraint.
-> If you underestimate data growth, you will pick the wrong database. If you pick the wrong database, you will rewrite the system in six months.

4. If you analyze one trade-off, analyze the impact on Functional Requirements. 
-> "Consistency vs. Availability" is an abstract concept.
-> You need to make a concrete choice. Example: "We will show users stale data to ensure the system survives a partition."

5. If you want to understand one thing about non-functional requirements (NFRs), stop listing their definitions 
-> Do not just list "Low Latency" or "High Availability."
-> Explain what they mean for this specific question. "Low" for a chat app is different from "Low" for a payment gateway.

Tools expire. Fundamental architecture patterns stay the same.

If you only know the tools, you aren't a System Designer.
You are just a walking README. md file.

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

How to create 10+ stories for your next Sr, Staff, EM behavioural interview from your last 2-3 projects using a 5-step process.

First think about your last 2-3 projects, which you have done in last 1-2 years. 

Keep a pen and paper ready to write down your stories.

Step # 1 

What is the first thing you do when you join a new project or company?

You need to ramp up yourself with: 
- the tool, 
- technology, 
- processes, 
- product

Write down: 
- how did you learn, 
- what challenges you faced, 
- what timelines were you tracking, 
- why learning them was important.

These are your learnability stories. 

Step # 2 

Now, when you work in a  project, what is the next thing you do?

You collaborate with stakeholders - product manager, your manager, colleagues, customers, QA, UX, BA etc.

These collaborations are about:
- understanding and prioritizing requirements 
- deciding which features to build, & which not
- disagreeing with them, aligning on a approach
- communicating with them to avoid confusions 

These are your stakeholder collaboration stories. 

Step # 3 

Now after requirements are finalized, what do you do next?

Design and implement the solution.

Think about:
- how did you dive deep to come up with solutions
- how did you use metrics, data to justify your choices
- how did you decide on tech stacks, design approaches
- did you have to choose quality vs speed of delivery
- did you innovate by making anything simple

These are your tech leadership stories.

Step # 4 

What is the next thing you do after above?

You take ownership and deliver the solution.

Think about:
- how did you deliver despite tight timelines
- how did you take complete ownership of delivery
- did you have to make tradeoffs between short and long term solutions?

These are your project delivery stories.

Step # 5

Once then project is delivered, what is the next thing you do?

You support the customers and continuously improve it.

Think about:
- the difficult discussions you have had with customers
- did you ever had to say NO to them, and why
- did you went above and beyond while supporting them

These are your customer focus stories. 

Bonus (Step # 6):

This is an overarching step while doing all the above steps.

- How did you help, and supported your colleagues?
- Did you mentor, coach them, if yes, what were results.


Thats all. 

If you took notes while reading this post, then congrats you are ready with 10+ stories that will cover 99% of behavioural interview questions.  

If you couldn't take notes, then save this post for later.

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

A 20-year EM or Senior engineer applies for a job.
But their resume says "14 years of experience."

This is a real (and flawed) strategy I'm seeing. 
Senior engineers and EMs are intentionally cutting their experience.

They assume they aren't prepared. 

They believe their stories and projects for 20 years aren't "good enough." 

They think "14 years" will make the interview easier.

This is a panic move. And it solves the wrong problem.

If you can't tell compelling stories about your 20-year career... You can't tell them for a 14-year career either.

The problem isn't the number. It's the story.

This fear comes from two root causes:
1. You became too comfortable in your job.
2. You lost motivation and stopped doing two things.

First, you stopped deeply reflecting on your projects and career stories.

Second, you stopped putting effort into upskilling.

By upskilling, I don't just mean the latest AI trend. 

I mean the core skills:
- Building a deep understanding of systems.
- Improving your leadership and behavioral skills.

When you suddenly decide to switch, you realize your knowledge is weak. Your stories are shallow. 

So you try to hide your experience instead of fixing the gap.

Don't be in that boat. Start working on this today:
1. Actively reflect on your career stories.
2. Spend time building your core technical and behavioral skills.

You don't need to hide your 20 years. 
You need to be able to explain them.

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

"We'll use events and CDC for loose coupling."
This statement sounds good in a design doc.

In reality, it's a top reason for "silent" production failures.

An upstream system (Service A) produces data. A downstream system (Service B) needs that data.

To avoid a "tight coupling," the engineer has Service B "listen" for changes from Service A.

Maybe Service B uses Change Data Capture (CDC) to stream changes from Service A's database. Or it just consumes from a generic event log.

Service A doesn't even know Service B exists. This feels like a win for loose coupling.

It's actually a time bomb.

And then reorg happens, team gets changed completely. 

The failure happens 3-6 months later.

The team for Service A changes their data contract. 
They rename a field. 
They refactor the code and stop producing a specific event.

Why? Because they forgot Service B was silently listening.

The dependency was implicit. 
It wasn't obvious in their code.

Service A's tests pass. They ship their change.

Weeks later, Service B breaks. The data is corrupt. The system is down in production. 

The damage is done, and it takes days to trace and fix the problem, which happened due to a change made a month ago.

That's why, stop relying on silent event streams for critical data flows.

Use an explicit command instead.

This doesn't mean it must be a synchronous API call. It can (and often should) still be an event.

But Service A must explicitly publish a well-defined event. 

The code in Service A should literally say: event_publisher.send("OrderProcessed_v1", data)

Now, the dependency is explicit.

When the Service A team refactors their code, they see this line. They can't forget it.

They are forced to think: "Who consumes OrderProcessed_v1? Oh, Service B. We are moving to v2, so we need to tell them."

This conversation happens during development. Not during a production fire.

Don't confuse "loose coupling" with "implicit dependencies."

One is a good design goal. 
The other is a production incident waiting to happen.

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

I recently helped a Senior Engineer land a critical role, when the odds were stacked against him.

He is on a work visa in the US. He comes from a country where the current geopolitical situation is little tricky. And the job market is brutal.

He was under pressure. He needed a win.

When we started, his experience was solid, but his delivery was hurting him.

His stories started with technology and ended with technology. This works for a junior engineer. It is a red flag for a Senior or Staff role.

The impact was missing.

We had to tear down his preparation and rebuild it.

We dug into his projects to find the real substance:
- How he handled disagreements with stakeholders.
- How he managed difficult customers.
- The trade-offs he made between quality and speed.
- How he justified complexity to his managers.
and many more...

We drilled his System Design (HLD) not just on "what" he built, but on the "why."
- Why that specific data store?
- How did he handle single points of failure?
- What were the architectural trade-offs?

The result?

He cracked a Senior Engineer role at a healthcare startup. It isn't a FAANG company. But it is a place where he creates real impact and feels valued.

Most people would stop there. They would relax.

He didn't.

Two weeks into the new job, he sent me the message in the screenshot below.

"My goal is not to be promoted. It is to act and think like a tech lead in my current role."

- He isn't asking for a promotion checklist.
- He isn't asking for a roadmap. 
- He isn't waiting for permission.

He wants to build the necessary skills to think like a Tech Lead in his current role. 

He understands that the title is just a side effect of the competence.

This is how you own your career.

While others wait for a sign that they are ready, engineers like him take action.

- They move fast. 
- They fix their gaps. 
- And they focus on the next important thing.

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

Everything is common sense, after someone explains it.

taplio

Chandra Shekhar Joshi's Linkedin Analytics

Chandra Shekhar Joshi

Performance ScoreChandra Shekhar Joshi's LinkedIn Performance

Chandra Shekhar Joshi's Best Posts (last 30 days)

Ready to grow your LinkedIn presence?

taplio

.css-6ytdur{background-image:linear-gradient(284deg, rgba(39, 43, 72, 1) 0%, rgba(33, 47, 183, 1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Chandra Shekhar Joshi's Linkedin Analytics

Chandra Shekhar Joshi

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);}Chandra Shekhar Joshi's LinkedIn Performance

Chandra Shekhar Joshi's Best Posts .css-1emoo84{font-size:var(--chakra-fontSizes-md);font-weight:400;color:var(--chakra-colors-gray-600);}(last 30 days)

Ready to grow your LinkedIn presence?

Chandra Shekhar Joshi's Linkedin Analytics

Performance ScoreChandra Shekhar Joshi's LinkedIn Performance

Chandra Shekhar Joshi's Best Posts (last 30 days)