What an AI/ML Fresher Salary Really Looks Like in Bangalore (2026)

There is a version of the Bangalore AI job market that exists mostly on social media. In that version, freshers walk out of college into twenty-lakh packages, everyone is a "GenAI engineer," and anybody earning less has simply not learned the right framework yet. Then there is the version that exists in offer letters, and the two are not close.

I have spent a long stretch of the past year paying attention to this gap — reading placement reports, going through job listings line by line, and talking to people at the point in their lives where the number on the offer letter feels like a verdict on them. Students, mostly. A few recruiters. A couple of founders who hire one or two junior engineers a year and think carefully about it.

What struck me first was how much shame there is around this topic. People will tell you their CGPA and their interview rounds happily, and go quiet about the number. That silence is exactly what allows the inflated version to circulate unchallenged, because the only figures spoken loudly are the outliers.

So this article is an attempt at the unglamorous version. What an AI or ML fresher in Bangalore realistically earns in 2026, why the range is so wide, what the money actually looks like after deductions, what it buys you in this city, and — the part I think matters most — what genuinely moves the number.

The short answer: Most AI/ML freshers in Bangalore in 2026 start somewhere between roughly ₹4 and ₹12 LPA, with service companies clustering at the bottom, product companies and funded startups in the middle, and a small number of top-tier offers well above it. In-hand is typically around 70–75% of CTC divided by twelve. Skills, internships and referrals shift the band far more than college name.

Table of Contents

• How I put this together, and what it can't tell you
• Why the range is so absurdly wide
• Salary bands by role
• Company by company: what to actually expect
• Startup vs MNC, honestly compared
• CTC vs in-hand: a worked example
• Ten conversations, including the disappointing ones
• Skills that actually increase your offer
• What keeps freshers stuck at 4–6 LPA
• How recruiters actually evaluate AI freshers
• What it costs to live in Bangalore in 2026
• The five-year growth picture
• Negotiating your first offer
• Frequently asked questions
• Key takeaways

How I Put This Together, and What It Can't Tell You

I want to be straightforward about method, because salary articles are usually vague about it and that vagueness is where the nonsense creeps in.

This piece draws on a few different things. Publicly visible job listings on LinkedIn, Naukri, Wellfound and company career pages, which sometimes state a band and more often reveal the level of the role. Salary-sharing platforms including Glassdoor, AmbitionBox and Levels.fyi, which are self-reported and skewed but still useful in aggregate. Engineering college placement disclosures, which are published but presented flatteringly. Conversations with recent graduates, recruiters and a handful of hiring managers. And a fair amount of cross-checking, because no single one of these sources is trustworthy on its own.

Now the limitations, which are significant.

Self-reported salary data has a selection problem: people who negotiated well report enthusiastically, people who took what they were offered often do not. Placement reports quote highest packages prominently and median packages quietly, if at all. Job listings frequently omit compensation entirely. And any figure attached to "AI engineer" covers roles that share almost nothing except the phrase — a person fine-tuning models and a person writing API glue around a vendor's model are both, on paper, AI engineers.

So the ranges in this article are working estimates for orientation, not benchmarks to hold an employer to. Before you make a decision on any of them, check current listings and salary-sharing communities yourself. Numbers in articles — including this one — age faster than the articles do.

A note on the profiles below: The conversations in this article are anonymized composite profiles drawn from recurring hiring experiences, recruiter discussions and publicly shared career stories across the Bangalore market. They illustrate real patterns rather than documenting specific individuals, and the salary figures attached to them are approximate context, not verified records.

Why the Range Is So Absurdly Wide

Ask what an AI fresher earns in Bangalore and the honest answer spans a factor of five. That sounds like evasion. It isn't — it reflects something real about how this market works, and understanding why is more useful than any single number.

The job title means almost nothing. This is the biggest factor and the one people underestimate. "AI Engineer" in a services company can mean integrating a vendor API into a client's workflow. In a product company it can mean owning a recommendation pipeline serving millions of requests. In a research team it can mean reading papers and running experiments for months. Same two words. Wildly different work, and therefore wildly different pay.

Hiring channel. Campus placement, off-campus application, referral and internship conversion produce different outcomes for identical candidates. Campus offers are standardised across a batch by design, which protects weaker candidates and caps stronger ones. Off-campus has no such ceiling and no such floor.

Company economics. A services firm bills your time to a client at a rate that constrains what it can pay you. A product company with high revenue per employee has more room. This isn't generosity or its absence; it's arithmetic, and it explains most of the gap between the two.

Timing. Someone who joined during an aggressive hiring quarter and someone who joined during a freeze got different offers for the same skills. Nobody controls this and it feels deeply unfair, because it is.

What you can prove. Two candidates with the same degree, one with a deployed project and an internship and one with six course certificates, are not the same candidate to a hiring manager, and the offers reflect that.

Here is the thing that surprised me most as I collected these stories. The variable that predicted the offer least reliably was the college. It mattered — I'm not going to pretend tier-1 institutes don't have an advantage, particularly for campus access to top product companies. But among people who went off-campus, which is most people, evidence of work predicted outcomes better than pedigree did.

Salary Bands by Role

The table below is my best consolidated estimate for entry-level roles in Bangalore in 2026. Read the "typical" column, not the top of the range — the top of every range is a small number of people, and building your expectations around them is how people end up rejecting reasonable offers.

A word on the "Prompt Engineer" row, because it's the one that changed most. Two years ago this was a genuinely hot standalone title. It is now, increasingly, a skill folded into other roles rather than a job by itself. If you are building a career plan around that title specifically, I would think carefully about it.

Company by Company: What to Actually Expect

Naming companies is where salary articles usually go wrong, so let me be careful. I am not publishing a table of what specific firms pay, because those numbers move, vary by team and level, and I would be presenting self-reported data as fact. What I can usefully describe are the categories, because the category tells you most of what you need to know.

Global product and platform companies

Google, Microsoft, Amazon, Adobe, NVIDIA, Intel, Qualcomm, SAP Labs. These sit at the top of the fresher range in Bangalore, often substantially. They hire relatively few freshers directly into AI-titled roles, preferring to hire strong software engineers and let specialisation happen internally. Compensation typically includes meaningful stock, which makes the headline CTC less comparable to a startup's. Access is heavily weighted toward tier-1 campus processes, competitive programming performance and internship conversion.

Indian product companies and unicorns

Flipkart, Swiggy, Razorpay, PhonePe, Meesho, Groww, Juspay, Freshworks. This is where a lot of the genuinely interesting applied AI work happens — recommendation, fraud, search, forecasting, logistics. Packages are strong without being top-of-market, the work is real, and hiring is more open to off-campus candidates with demonstrable ability than the previous category. If you had asked me where a determined tier-2 or tier-3 graduate has the best odds of a strong first offer, I would point here.

Analytics and AI consultancies

Tiger Analytics, Fractal Analytics, Mu Sigma, MathCo, and similar. These hire freshers in volume, which is a genuine advantage when you are trying to break in. Pay sits in the middle band. The work is client-driven, which means exposure to many business problems and sometimes less depth on any one of them. Several people I spoke to described these as excellent first jobs and less good third jobs.

IT services

TCS, Infosys, Wipro, Cognizant, Capgemini, LTIMindtree, Accenture. The largest hirers of freshers by a wide margin, and the bottom of the salary range. Standard entry packages sit low, though most have digital or AI-specialist tracks with meaningfully higher bands that you can test into. I want to be fair here, because these companies get dismissed carelessly: for a candidate with no internship, no referrals and no portfolio, a services offer is a job, an income and a foot in the door, and plenty of strong engineers started exactly there. The risk is not the starting salary. The risk is staying long enough on maintenance work that you have nothing to show at your first switch.

AI-first startups

Observe.AI and the long tail of smaller, less visible companies. Enormous variance — some pay competitively with product companies, some pay well below market and offer equity as compensation for the gap. This is the category where due diligence matters most, and where the difference between a career-making first job and a wasted two years is widest.

Hardware, R&D and captive centres

Bosch, and the India R&D arms of various multinationals. Often overlooked by students chasing AI branding, frequently offering solid pay, real engineering problems and better working hours than the equivalent startup. Worth applying to.

Startup vs MNC, Honestly Compared

My honest read, having listened to people on both sides: the startup-versus-MNC debate is usually framed as a personality question when it's really a specificity question. A good startup beats a mediocre MNC team. A good MNC team beats a directionless startup. Ask about the specific team, not the category.

CTC vs In-Hand: A Worked Example

This is the section I wish someone had shown me. A great deal of disappointment in the first month of a job comes from not understanding what CTC means.

Take a ₹12 LPA offer at a product company. A plausible structure:

• Fixed base: ₹10,20,000 — the part your monthly salary is calculated from
• Performance variable: ₹1,00,000 — paid annually, contingent on rating and company performance
• Joining bonus: ₹50,000 — one-time, often clawed back if you leave within a year
• Employer PF contribution: ₹21,600 — real money, but it goes to your PF account, not your bank
• Gratuity provision: ₹8,400 — only payable after five years of service in most cases

Add it up and it says twelve lakh on the letter. But the joining bonus arrives once. The variable arrives once a year, if it arrives in full. The PF and gratuity components never appear in your account as spendable cash.

From the ₹10.2 lakh base, your monthly gross is about ₹85,000. Subtract your own PF contribution of roughly ₹1,800, professional tax of ₹200, and income tax under the current regime after standard deduction — which at this income level is modest but not zero. Monthly in-hand realistically lands somewhere around ₹72,000 to ₹78,000.

So "twelve lakhs" is, month to month, about seventy-five thousand rupees. That's still a good salary for a fresher in Bangalore. It is simply not one lakh a month, which is what a lot of people mentally divide it into.

Ask for the detailed salary structure before you compare two offers. A ₹12 LPA offer with a ₹10.2L base beats a ₹13 LPA offer with a ₹9L base and a large variable, and you cannot see that from the headline.

Ten Conversations, Including the Disappointing Ones

These are the composite profiles described earlier. I've deliberately included the ones that didn't work out, because salary articles have a habit of quietly filtering those out and that filtering is precisely what distorts everyone's expectations.

1. Rahul — BE CSE, tier-3 college, startup, ₹8.5 LPA

Went in expecting fifteen after months of YouTube videos about AI salaries. Was offered eight and a half. Took two weeks to accept it and described that fortnight as "the most miserable of my life, over nothing."

A year later he is unrecognisable technically — he owns a deployed retrieval pipeline and has been on-call for it. "I was upset about six lakhs of difference that didn't exist. The fifteen was never on the table for me. I just believed it was because a video told me."

2. Sneha — MCA, tier-2, analytics consultancy, ₹6.8 LPA

Applied to over a hundred and fifty roles; the consultancy was the only one that ran an open, structured process she could actually enter. Mid-band pay, heavy client work, three different industries in eighteen months.

Her learning: "Everyone told me consultancies aren't real AI. My last project was a forecasting model that a client actually uses. I don't know what would be more real than that."

3. Arjun — B.Tech tier-1, product company, ₹19 LPA

Campus offer, internship converted. The profile everyone points at. What he was clear about: the internship did the work, not the degree. "Six of us interned. Four converted. The two who didn't had the same college and the same CGPA as me."

4. Priya — B.Sc then self-taught, small startup, ₹4.5 LPA

The lowest offer in this set and, in some ways, the most interesting story. No engineering degree, no campus process, no referrals. She built a document-extraction tool for a local logistics firm as a freelance project and used it as her entire portfolio.

Four and a half lakhs is not a comfortable salary in Bangalore, and she was candid that the first year involved a shared room and careful budgeting. She switched after eighteen months and roughly doubled it.

Her learning: "The first offer isn't your salary. It's your entry fee."

5. Karthik — B.Tech tier-2, IT services, ₹4.2 LPA, later moved

Joined a services company through campus and spent fourteen months on work he described as "ticket-closing with a machine learning label on the project name." No deployment, no ownership, no new skills.

What saved him was rebuilding a portfolio in his evenings and switching. The switch took him to a mid-size product company at more than double the pay.

His learning: "The salary wasn't the problem. The problem was that after a year I had nothing to talk about in an interview. That's what nearly trapped me." If this sounds familiar, our piece on portfolio projects that actually got interviews covers the rebuild in more detail.

6. Meera — M.Tech, computer vision role, ₹13 LPA

One of the few people in these conversations for whom the postgraduate degree was straightforwardly load-bearing. The role required reading papers and reimplementing methods. "They asked me to derive things. If I'd learned this from a course I'd have been finished in ten minutes."

She was also the most cautious about generalising from her own case: "This is one job. Most AI jobs are not this job."

7. Farhan — career switcher from testing, ₹9 LPA after 14 months

Four years in manual QA, wanted out. Studied evenings and weekends for over a year. His advantage turned out to be one he hadn't valued: he understood how software breaks in production, which most fresh graduates do not.

His learning: "I applied as a fresher and got fresher offers. I should have applied as an experienced engineer moving into ML. Same person, different framing, probably different number."

8. Divya — remote role for a Bangalore startup, ₹7.5 LPA

Works from a tier-2 city for a Bangalore-headquartered company. Lower nominal salary than her in-office peers, dramatically better in real terms because her rent is a quarter of theirs.

Her learning: "People compare CTC across cities like it means the same thing. My friend in Bangalore earns more than me and saves less."

9. Nikhil — three offers, negotiated one, ₹11 LPA

The only person in this set who negotiated meaningfully, and the details are instructive. He had two competing offers, mentioned this once, politely, in an email, and asked whether there was flexibility. The company raised the base by about a lakh and added a joining bonus.

His learning: "I was terrified they'd withdraw. They didn't even sound surprised. I think I was the only fresher who asked."

10. Anonymous — offer withdrawn, then ₹5.5 LPA six months later

Included because it happens and nobody writes about it. Had a strong offer from a funded startup; the company restructured before the joining date and the offer was rescinded three weeks out. Six months of applying followed, ending in a considerably smaller role.

Their learning: "Nothing I did caused it and nothing I could have done would have prevented it. I stopped treating my salary as a score after that."

Skills That Actually Increase Your Offer

Sorted roughly by how much difference each one made in the conversations I had, rather than by how often it appears on a syllabus.

The ones that consistently moved offers

• SQL, genuinely fluent. The most common regret, by a distance. Unglamorous, tested constantly, assumed rather than taught.
• Python that another engineer can read. Not just Python that runs. The gap between those two is where a lot of junior candidates lose credibility.
• Deployment. Docker, one cloud provider, an endpoint that stays up, basic logging and monitoring. A model in a notebook is homework; a model behind an API is engineering, and interviewers can tell instantly which one you've done.
• One deep-learning framework, used properly. PyTorch or TensorFlow. Depth in one beats familiarity with both.
• RAG, evaluation and vector databases. Currently the clearest differentiator for GenAI-adjacent roles. Specifically the evaluation part — plenty of candidates can build a retrieval demo, far fewer can explain how they measured whether it was any good.
• Data structures and algorithms. Still gates the first round at many companies, including for ML roles. You do not need to be a competitive programmer. You do need to not fail the screen.
• An internship. Not a skill, but it shifts offers more than most skills do.

Useful, but rarely decisive on their own

• Git and Linux. Expected rather than rewarded. Their absence hurts more than their presence helps.
• Kaggle. Good for demonstrating iteration on messy data. Less impressive than a deployed system, because competitions never test deployment.
• Open-source contributions. Strong signal when real — a merged pull request means somebody reviewed your code and accepted it.
• Research papers. Decisive for research roles, largely irrelevant for applied ones.
• System design basics. Rarely tested at fresher level, but knowing why you'd cache something reads as maturity.

The one nobody lists and everyone mentions

Communication. Every recruiter I spoke to raised it unprompted, usually with some frustration. Being able to explain what you built, why you chose that approach, and where it falls short — in two minutes, out loud, without notes — separated candidates more sharply than any technical skill. It is also the most trainable thing on this list, and almost nobody practises it.

What Keeps Freshers Stuck at 4–6 LPA

These patterns came up repeatedly, and none of them are about intelligence.

Certificates instead of artifacts. Fourteen completed courses and nothing built is an immediately recognisable profile, and not a flattering one. It reads as consumption rather than production.

The same five tutorial projects. Digit classifier, sentiment analysis on a standard dataset, movie recommender, churn prediction, chatbot. Reviewers have seen these thousands of times. They convey nothing because everyone has them.

Portfolios copied from GitHub. More common than you'd hope, and trivially caught. One clarifying question about a design choice ends the interview.

A resume that lists skills instead of evidence. "Python, ML, DL, NLP, AI" tells a recruiter nothing. One line describing what you built and what happened tells them everything. We covered the rebuild process in this piece on fixing a resume that kept getting rejected.

No interview preparation. Knowing the material and performing under time pressure are different skills. Structured practice closes that gap faster than more learning does.

Applying at volume with one generic resume. Two hundred untargeted applications reliably produce silence. Twenty tailored ones outperform them and take less total effort.

An empty LinkedIn. Often the second thing checked after your resume. A profile reading "aspiring data scientist" and nothing else is a wasted slot.

Waiting until you feel ready. That feeling does not arrive. Several people described losing six to nine months to it, and every one of them said the rejections taught them more than the extra courses did.

How Recruiters Actually Evaluate AI Freshers

Compressed from what hiring managers described, roughly in sequence.

Resume screen — often under a minute. They look for evidence of building, an internship if one exists, and a working link. A broken demo link costs more than untidy code, because it suggests you didn't check.

Coding round. Frequently standard algorithmic problems, even for ML roles. This filter exists partly because it's cheap to run at scale, not because it perfectly predicts performance.

Machine learning fundamentals. Bias-variance, overfitting, why a metric is or isn't appropriate, what happens under class imbalance. Interviewers are checking whether your understanding survives one follow-up question.

Statistics and SQL. More heavily weighted than students expect, particularly for data science roles.

The project discussion. Usually the deciding round. They will pick something on your resume and dig. What did you try first? Why did that fail? What would you do differently? What doesn't your solution handle? Candidates who can answer the last question honestly do noticeably better than candidates who claim their project has no limitations.

Deployment and practical judgement. Would you ship this? What breaks at ten times the traffic? How would you know if it degraded?

Curiosity. Mentioned surprisingly often. Managers hiring freshers know they're buying potential, and someone who has clearly gone digging into something on their own is a better bet than someone who has only completed what was assigned.

What It Costs to Live in Bangalore in 2026

Salary means nothing without this half of the equation. Rough monthly estimates for a single person, which vary sharply by area — the same flat costs very differently in Koramangala and in Electronic City.

Applied to three salary levels:

At ₹6 LPA (roughly ₹42,000 in hand): a frugal life works. PG, home-cooked or mess food, public transport, occasional outings. You can save perhaps eight to twelve thousand a month if you're disciplined. It is genuinely tight if you're also sending money home.

At ₹10 LPA (roughly ₹62,000 in hand): comfortable without being lavish. Your own room, some eating out, cabs when it rains, and realistic savings of fifteen to twenty-five thousand a month.

At ₹15 LPA (roughly ₹92,000 in hand): a good standard of living plus meaningful saving — thirty-five to fifty thousand a month if lifestyle inflation doesn't absorb it, which it very often does.

Whatever the level, build a small emergency fund early. Three to six months of expenses. Several people in these conversations discovered why during a layoff or a rescinded offer, and the ones who had it described the difference as enormous.

The Five-Year Growth Picture

Two caveats that matter more than the table.

First, those ranges are wide because outcomes diverge fast. Two people who joined the same week at the same salary routinely sit two bands apart by year three, and the difference is almost always what they were allowed to own rather than how long they stayed.

Second, the growth is not automatic. Years accumulate whether or not skills do. The engineers whose compensation climbed steeply were, without exception, the ones who kept taking on things slightly beyond their current level.

Negotiating Your First Offer

Freshers negotiate far less than they could, mostly out of fear that asking will cost them the offer. In practice, a polite, single, reasoned ask almost never does.

What actually gives you leverage: a competing offer, a specific skill the team said it needed, an internship at the same company, or a rare specialisation. What doesn't: what your friend earns, what a YouTube video said the market pays, or your own financial needs, however real.

If base pay is fixed — and on campus offers it usually genuinely is — other things sometimes aren't. Joining bonus. Relocation allowance. An earlier first review. A written commitment about the team or project you'll join, which at fresher level is arguably worth more than the money.

Ask once. Ask specifically. Accept the answer gracefully either way. Nikhil's line stayed with me: "I think I was the only fresher who asked."

Frequently Asked Questions

Is 6 LPA a good salary for an AI fresher in Bangalore?

It is workable rather than comfortable. Around ₹45,000 in hand covers a shared flat or PG, food, transport and a modest social life, with a little left to save. It does not stretch far if you're supporting family. At this stage, what the role teaches you is worth more over three years than a lakh or two of starting CTC.

Can a tier-3 college student get a 12 LPA AI job in Bangalore?

It happens, but rarely through campus placement. The route is off-campus: strong deployed projects, a known internship, referrals or open-source work that gets your application opened. A more realistic first target is the mid single digits, followed by a substantial jump at the first switch.

Do AI engineers earn more than software engineers as freshers?

Usually not by much at entry level, since both need supervision. The premium tends to appear at two to five years, when an ML engineer who can ship and maintain production systems becomes hard to replace.

What is the real difference between CTC and in-hand?

CTC includes employer PF, gratuity provisions, insurance, one-time joining bonuses and annual variable pay — none of which reach your account monthly. In-hand is roughly 70–75% of CTC divided by twelve for entry-level packages, depending on structure.

Can freshers negotiate salary in Bangalore?

Yes, more often off-campus than on. Leverage comes from competing offers, scarce skills or an internship at the company. Where base is fixed, joining bonuses and review timing sometimes aren't.

Should I take a startup offer or an MNC offer?

Ask about the specific team rather than the category. Find out who will review your code, how long the funding runway is, and what the last three engineering hires are working on now. Those answers predict your next two years better than the CTC does.

How much do AI internships pay in Bangalore?

Anywhere from nothing to over a lakh a month. Large product companies and funded startups pay well; small companies often pay little. The conversion opportunity and the workplace experience usually matter more than the stipend at this stage.

Are ESOPs worth anything for a fresher?

Treat them as a lottery ticket. Evaluate the offer on cash and consider equity a bonus. If you want to assess it seriously, ask about vesting, cliff, strike price, last valuation and buyback history.

How important is CGPA?

Mainly as an early filter — many processes apply a cutoff below which nothing is read. Above it, small differences rarely decide anything, and product companies and startups weigh projects and interviews far more heavily.

Which skills raise a fresher offer fastest?

Solid Python, fluent SQL, one deep-learning framework used properly, and the ability to deploy and monitor something. Practical RAG and evaluation experience is currently scarce enough to shift a band on its own.

Do I need a master's degree for AI jobs?

For research roles and some computer vision or NLP positions, often yes. For applied engineering, frequently not. We looked at this in detail in this piece on whether a master's is required for AI jobs in India.

Is 4 LPA too low to accept?

It's low, and in Bangalore it's tight. Whether to accept depends on the alternative. A low-paying role with real ownership and a mentor can be worth more than six months of unemployment. A low-paying role with neither is worth leaving as soon as you have something better.

How long should I stay in my first job?

Long enough to have shipped something you can describe in detail — usually twelve to twenty-four months. Leaving before that often means leaving without a story. Staying much beyond it on stagnant work has its own cost.

Do certifications increase salary?

Rarely on their own. They provide structure while learning. What converts a course into a hiring signal is what you build afterwards.

Is remote work paid less?

Often nominally, yes, though the real-terms comparison can favour remote heavily if you're living somewhere cheaper. Compare savings, not CTC.

Are AI jobs in Bangalore still growing in 2026?

Demand for people who can build and maintain production AI systems remains strong. Demand for people who can only prompt a model has weakened noticeably as that skill has become common. The market has become more discriminating rather than smaller — a shift we covered in this piece on what changed for entry-level AI jobs.

What if my offer is rescinded?

It happens, it is usually nothing to do with you, and it is more common than public discussion suggests. Keep applying while you wait for a joining date, and keep an emergency fund.

Should I do free work to get experience?

A single small unpaid project for a real user can be worth it as a portfolio piece. Ongoing unpaid work for a company that can afford to pay is exploitation, and it does not lead anywhere.

Does location within Bangalore affect my real salary?

Considerably. Rent varies by a factor of two or three across areas, and a long commute costs money and hours. Living near the office at a slightly higher rent is often financially neutral and materially better for your life.

What single thing would raise my offer most?

One deployed, documented project that you can defend under questioning for thirty minutes. Nearly every person in these conversations pointed at something like it, and almost nobody pointed at another certificate.

Key Takeaways

• Most AI/ML freshers in Bangalore start between roughly ₹4 and ₹12 LPA; the viral numbers are real but rare.
• The job title tells you almost nothing — the same words cover very different work at very different pay.
• In-hand is typically 70–75% of CTC divided by twelve. Always ask for the detailed structure.
• Company category predicts pay better than your college does: services at the bottom, consultancies mid, product companies and top-tier firms above.
• Evaluate startups on the specific team, mentorship and runway. Treat ESOPs as a bonus, never as salary.
• SQL, readable Python, deployment and one framework used properly move offers more than any certificate.
• Communication is the most-mentioned and least-practised differentiator.
• Bangalore living costs mean a ₹6 LPA salary is workable and a ₹15 LPA salary is comfortable — compare savings, not CTC.
• The first offer is an entry fee, not a verdict. The largest jumps tend to come at the first switch.
• Check Glassdoor, AmbitionBox, Levels.fyi and current listings yourself. Numbers in articles age badly.

The Part Worth Remembering

The thing that stayed with me from all of these conversations wasn't a number. It was how differently people talked about their first salary a year or two later compared to how they talked about it at the time.

At the time, it was everything. It was a score, a comparison against classmates, evidence of worth. A year later, almost nobody mentioned it unprompted. What they talked about instead was what they had been allowed to build, whether anyone had reviewed their work, and whether they had learned something they couldn't have learned alone.

That isn't a reason to accept an unfair offer. Know your worth, ask when you can, and walk away from roles that will waste your time. But the leverage you're looking for — the thing that actually changes the number — isn't available at the offer stage. It's built in the months before it, and then again in the year after.

Rahul spent two weeks miserable over six lakhs that were never on the table. A year later he owns a system in production. He is going to earn considerably more than the figure he was chasing, and none of it will come from the offer he agonised over.

Build the thing. The number follows.