Thoughts on how I predict AI will play out in 2026 - especially in the field of software engineering.

From my vantage point and the conclusion of the tinkering around I've been doing with it.

1. The "AI bubble" will not burst in 2026. Likely will in 2027 2. Consumer hardware will get sold on AI capabilities. Like gamers comparing GPU flops, photographers comparing megapixels - professionals will start buying hardware based on ability to run mid sized coding models and ComfyUI workflows locally. Apple has a lead on this.

Today I made my Uber driver wait for 2 min as I was getting out of my house. And suddenly I received a notification from Uber "I am facing the threat of murder"

Strap in for a thrilling story 🧵👇 A chill ran down my spine.

It is Delhi afterall. Anything can happen.

Is he threatening to murder me because I made him wait and mistyped it?

Are people on the street threatening to murder him because he is blocking the road?

All sorts of thoughts raced in my head.

Exactly one year back I made the fateful decision to try dipping my toes in a part of the tech industry I have not remotely been close to - datacenters.

Last 10 years I have built startups and led engineering teams largely around two things - edtech and consumer mobile apps. A wonderful thing about bigtech hiring process is that often the hiring is generic to a role, and then you get to pick among various teams that have open headcount. (no other way hiring would scale to this size).

Felt like a great opportunity to explore something new.

This post is written by someone who is IIT+IIM, ex CIO (Chief Information Officer, equivalent to CTO in some orgs) and multiple leadership roles.

A lot of business leaders have this level of technical understanding. Hard to build technical things under them. The problem is not only that they wouldn't have the understanding that the "special laptop" is just having Wifi 6 or higher throughput network card, they won't even have the patience for this to be explained to them, and shout at you thinking you're trying to hoodwink them.

It is interesting to see the various ways people approach the concept of "growth", especially since I have to discuss this topic as a manager, and also when mentoring people

People always want to "grow", but aren't often aware whether how to exactly do that In a formal manager relationship I often encounter a situation where someone wants to be given task they have never done before, and want to prove they can do it, and it's the easiest idea in most people's minds on how to "grow".

It is not wrong. Growing is doing new things.

How to build an efficient engineering team

1. Instead of paying at 90th percentile, pay at 50th percentile. In this job market you’ll still be able to hire

2. You’ll get SDE2 at 20L instead of 30L, Seniors at 30L instead of 45L and Staff at 50L instead of 70L 3. Hire EMs because the cheaper SDEs are both less productive and less motivated and don’t take ownership. So you need EMs to “drive” and “coach” them

4. Instead of hiring senior PMs, hire APMs from altMBA bootcamps. You’ll easily get APMs at 15L

I keep reading about Indian tourists causing a scene (like the Kenya safari tweet thread recently), and got to experience it first hand today

Went to an Indian restaurant in Vienna today after 10+ days of no Indian food, and it looked beautiful from the outside So it is a Mughal-themed restaurant (Moghul was literally in the name) and had some really nice large paintings of Mughal kings inside

Was talking to the owner - a 30 yr old Punjabi chap, born and brought up in Vienna. His parents running this place since 37 years (now he does)

Forget the typical “roadmaps” and other drivel the “get into tech” cottage industry on YouTube by a bunch of FAANG SDE1s has generated.

If you want to be an SDE2/SDE3 at a typical growth stage product engineering team (eg: Zepto, Upstox, Purplle, LivSpace) would love to get… …here’s a few well articulated little assignments that’s gonna get you right there in the sweet spot of “highly desirable growth stage engineer” zone

The main pillars are
- structuring code neatly
- strong concurrency fundamentals
- can model non-trivial db schema
- utils 😜

1. All engineering teams need some "slack time" (not the chatting app)

2. All engineering teams should be slightly understaffed and perennially non-empty list of things to do

The biggest Engineering Management learning is to reconcile these two truths of work. There is this famous book by Tom Demarco called "Slack" which is great read on why teams need some 'slack time' - i.e. if the team's capacity is to produce 200 units of work per week, then only plan for 175 units of tasks every week, and kee 25 units free to pick up tasks that come along the way.

Things break.
Suddenly things get re-prioritized.
You encounter weird corner cases working on something, and it takes more time to close.
Shit happens.

Always have some spare time to keep bandwidth for that.

Story time. This is the saga of someone famous. See if you can guess it.

-- Part 1 - The Teacher --

So there was this guy who loved teaching people. It didn’t really start off as being a professor in a classroom or something. More like, helped younger cousin with Carnot cycle during his 12th grade physics exam, or opens up a paper an pen to explain things with diagrams to co-workers often.

Declarative UI and one-way flow of state is something that invariably developers seem to prefer and all frontend engineering platforms have been progressively moving towards it.

The problem though lies that most developers starting out are not coached properly for it. Let's try to understand the problem here.

Classically how UI is represented and how it triggers logic is a bit like this

UI interacts with some "controller" like layer which in turn interacts with local DBs or APIs.

That's pretty much the basic 'structure' to this model.

How to travel if you're broke?

A thread 🧵 👇 1. Sell your Jaguar

Imagine you build a hyperlocal delivery app (like Swiggy/Delivery Hero/Uber Eats etc).

The top level API response for a user typically is all stores around them, typically paginated (infinite scroll etc) with info like name, rating, a thumbnail, estimated ETA etc At the top level response the information you show is a aggregated/derived from many deeper levels. For example -

Rating: this is avg cumulative rating, that is affected as more ratings come in

ETA: this might depend on time of day / traffic conditions and user's location

So I find myself taking 8-10 interviews a week right now, and whenever you're in a hiring storm - out comes a lot of insights into the interview process, and potential learnings for people (both on hiring and job-searching sides).

So thus starts a thread. This is my 3rd time in the middle of large, long drawn, hiring "campaign" for a big team about to expand into a large team. (Done it twice before at Zomato and Target - where you have a 50-60 headcount to fill, over a period of 2-3 months, across the usual job roles).

Cooking black zip ties today. For something called "squid ink spaghetti" turned out exactly as advertised.

When some of my juniors, and former students ask my about career dilemmas, often I hear the equivalent of "But I am at L4 here, and that offer is L3 only, should I try to get L5 in that other place", while the 3 companies are wildly different

Let's break it down. So the first thing that is important is to understand what these "L" levels even mean. And for that we will have to look at a generalised, most common, story of how the tech team pyramid grows in a company.

It often starts out with the 'Tech Cofounder' writing code.

Consulting rates:

$50/hr: I'll do what you ask me to do

$500/hr: I'll solve your problem somehow

$5000/hr: I'll tell you what to do (and slightly help in doing it)

$50000/hr: I'll tell you *why* you should do things differently Job Titles

$50/hr: Freelancer

$500/hr: Agency

$5000/hr: Strategy Consultant

$50000/hr: Management Consultant for Transformation

No amount of data science can prepare you for the kind of insights I just got from a guy I know works at an edtech company.

"Our founder proposed to our co-founder on stage and that caused a significant inbound volume and spike on course purchases this week". New Valentine's Day growth hack dropped - get your founders hitched to each other. 🤪

A thread of some informative videos on careers in software engineering I made over the last few months on @scaler_official's YouTube channel.

It is a series called "Refactor". Jobs at Startups vs MNCs

- what to expect in terms of roles
- compensation structure differences
- how to prepare for each of them

In the last few years some of the biggest mobile apps(in terms of users) have been built out of India. That has made this fertile ground for homegrown "app PMs".

And last year they all shuffled around all the new VC funded companies. Lo and behold all apps are the same now All buttons have mIcRoInTeRaCtIoNs now. Small multicoloured polka dots exploding from every little thing that you touch.

But no one has learnt to consistently put negative/positive CTAs always on same side on every dialog of the app.

Same PMs, same mistakes :)

Saw a Scorpio with Nazi emblem in Bangalore today. Things are getting wild out here.

(Cannot really give the 'hindu swastika' benefit of doubt because of the 45° angle, and red and black colors). People trying to pull a Hanlon's Razor here - you know you are part of the problem.

This is not an 'innocent' mistake for sure. It is a circular sticker, so the person had to have 'deliberately' put it up at 45° angle, which means they very well know what it is.