Sahil Gupta

SSE @HSBC • Sharing what I'm learning about backend and AI systems

[Sahil Gupta @guptasahil7]

been on this app since 2017 and posted like 40 times.
turns out that's not how it works.

fixing that. starting with the stuff I've actually written 👇

[Sahil Gupta @guptasahil7]

A senior gave me this analogy about a year ago and it stuck with me he said...

Treat your job like a bank account.

delivering something high impact under a tight deadline → credit

taking ownership when things go
wrong → credit

helping the team when they need you → credit

emergency leave, a mistake, missing a commitment → debit

You’ll have debits... that's normal.

Just keep building enough trust that your balance stays positive.

[Sahil Gupta @guptasahil7]

Caught Under the Mangosteen Tree yesterday 🎭

What a great evening

Multiple love stories exploring sensitive topics like class difference, interfaith relationships, incomplete love and non-binary marriages all told in such a simple and loving way

Maybe that's what I like about theatre... you watch the same story but everyone walks away with something slightly different

[Sahil Gupta @guptasahil7]

@claudesfun Hope not a scam

[Sahil Gupta @guptasahil7]

I see people talking about building new businesses all the time

Not enough about the businesses that already exist and will eventually need someone to take them over

came across this interesting report from @McKinsey on the wave of small businesses that will change hands over the next decade

They estimate ~6M businesses could go through ownership transitions by 2035... with up to $5T in enterprise value at stake

That creates a completely different set of problems to solve

The interesting part is that most of these businesses aren't necessarily failing, lot of them just don't have a clear path to a new owner

There’s something interesting here.

https://t.co/qKPSpW6B2W

[Sahil Gupta @guptasahil7]

Yeah… a couple of these are definitely me 😅

[Hitesh Choudhary @Hiteshdotcom]

A must watch for all devs 😌

[Sahil Gupta @guptasahil7]

Came across this interview with @dhh

This one is worth 25 mins if you are a software engineer

Some really interesting thoughts on AI, programming, building software and where the craft is heading

Worth a read if you're into building stuff

https://t.co/zMXWv410KV

[Dmitrii Kovanikov @ChShersh]

You can have bad days.

And then you have even worse days.

And you realise bad days were good days.

[Sahil Gupta @guptasahil7]

You learn a system much faster when you stop looking at your service in isolation

Don't just know what your service does.... understand what talks to it, what it talks to, where the data comes from and where it goes

And more importantly...understand why it was designed that way

Why this service owns this data.
Why this component is separate.
Why this flow exists.

That's something tutorials can't really teach you

[Sahil Gupta @guptasahil7]

Did an improv workshop yesterday with a bunch of strangers

Basically acting games, random scenes and a lot of making things up on the spot 😅

Was genuinely fun....if anyone in Pune wants to try one... DM me. I can share the details

[Sahil Gupta @guptasahil7]

@mirajobs Yeah...i usually start with the trace if I have it...... it cuts down the search space quite a bit before i even get into the logs

[Sahil Gupta @guptasahil7]

One thing I have learnt debugging distributed systems: don't treat logs as the starting point.

When a request crosses multiple services, searching for ERROR usually gives you a pile of unrelated noise

Start with one request....follow its trace across the services and find where the behaviour or latency actually diverges

Then use the logs around that span to understand what happened

Trace gives you the path. logs give you the context

The trick isn't finding more logs. It's narrowing the search space until you have something worth debugging.

[Sahil Gupta @guptasahil7]

@PuneDistrict

[Sahil Gupta @guptasahil7]

Go to decathlon --> buy a few things --> convince myself i'll finally start going to gym --> never actually go

[Sahil Gupta @guptasahil7]

This is basic stuff, but I have seen plenty of junior devs get this wrong

They change the contract assuming they only need to worry about their own service

Think of an API contract like a published document.

Once people start building against it, your API isn't really yours anymore....they are depending on the shape of your requests, responses, fields and even the behaviour

So whenever you want to change something, ask one question:

“Can I make this change without breaking what I've already promised ?”

If yes, keep the same version.

If no, you need a new version.

Adding a new optional field? Probably fine

Removing a field that consumers use? Breaking change

Changing what an existing field means? Breaking change

You don't version an API because you've made some changes

You version it when you've changed the contract

That's the mental model you can use

[Sahil Gupta @guptasahil7]

Designing API contracts nicely is the key to inter-team collaboration

I have seen many people start with the code, when they should start with the contract

A good contract answers three things:
1. what goes in
2. what comes out
3. what fails and how.

Once the contract is clear, both teams can work independently. They know what they need to send and what they can expect in return. You don't have to wait for the other team to finish their implementation before starting yours.

A contract is a promise. Once you publish it, you can't quietly break it.

[Sahil Gupta @guptasahil7]

And finally Part 4 of "AI Fundamentals, Stupid Simple" 😅

This one covers Transformers, Attention, Multi-head Attention and KV Cache..... basically what happens after we turn text into vectors

I have tried to keep this entire series as intuitive as possible and build things one brick at a time rather than throwing definitions and equations at you

That's it for the series.

Hope it helps someone build a better mental model of what's actually happening inside an LLM.

https://t.co/0kWQ81Jlmm

[Sahil Gupta @guptasahil7]

I’m a Senior Software Engineer at HSBC, working in the Prime Brokerage domain, where we build systems that help hedge funds trade by providing financing and managing their collateral.

On the side, I’m building https://t.co/B2aMDPv0F5, an AI-powered due diligence platform for people evaluating small business acquisitions.

just put myself on the map: https://t.co/1nn6yR9gx8
https://t.co/vsYwhOnhpG

[Darya Chachura @daryachach]

I’m 22.

I design and build my own mobile apps and do the marketing for them.

From Belarus, based in Poland 🇵🇱

Looking to connect with more developers, designers and marketers.

[Ryo Lu @ryolu_]

efficiency, productivity, speed

we keep worshipping these words like they are what we live for

ship faster
merge more
manage more agents
compress every loop
remove every pause
turn every hour into output

but what is the point of endless production if there is no time left to think deeply?

not react
not optimize
not summarize
not clear the queue

think

to sit with a question long enough that your real intention starts to appear
to find the purpose underneath the motion
to ask whether the thing being accelerated should exist at all

ai could be a new canvas

a way to make strange things
personal things
impossible things
tools, worlds, poems, games, interfaces, films, languages, rituals
things that let more people touch the shape of their own imagination

but instead, so much of it is becoming slop factories

more content
more funnels
more tickets
more fake work pretending to be value

you can merge 2000 prs
you can manage 500 agents
you can wake up to dashboards proving that the machine kept moving while you slept

but what is the point if you only sleep 5 hours a day?

what is the point if you stop talking to other humans?
stop learning about the past?
stop playing?
stop wandering?
stop being surprised?

what is the point if your whole life becomes optimizing the system that is consuming you?

speed is not meaning
scale is not wisdom
automation is not freedom if it only makes everyone run faster

and where are we running?

toward what?

a world where every person becomes a manager of machines, and every machine produces more things no one had the time to truly want?

the danger is not that ai makes us lazy

the danger is that ai makes us endlessly busy

that it gives us infinite production before we have found intention
infinite execution before we have developed taste
infinite motion before we have learned how to be still

maybe the real frontier is not more speed

maybe it is discernment

knowing what not to make
knowing when to stop
knowing when the most productive thing is to sleep, read history, call a friend, play with a cat, walk without tracking it, or stare at the ceiling until the false urgency dissolves

tools should give us more life
not turn life into a tool

if ai is going to matter, it should not just help us produce more

it should help us become more human
more imaginative
more free

otherwise, it’s the same old factory
just without humans

[Sahil Gupta @guptasahil7]

Managing DB connections at scale is a nightmare

Just got out of one and yeah… i can confirm 😭

Everything looks green until one day suddenly you hit a roadblock

Opening a connection is simple....the tricky part is knowing how many you actually need, when to create them, when to release them and what happens when everything starts piling up

Some problems only make sense once you’ve personally been burned by them

[Sahil Gupta @guptasahil7]

Heard a lot of people talking about this book lately, so I asked my friend to bring it over

Turns out, he already had it and he didn't quite like it 😅

So let’s find out for ourselves

Getting my hands on it now...have you read it? what did you think?