Somewhere in India tonight, a father is signing an education loan for his child's engineering admission. The brochure shows a glass building, a foreign campus tie-up, and the words "100% placement." He believes he is buying safety.
He is buying a product whose price has tripled in a decade while its output has not moved at all.
15 lakh
Engineering graduates produced in India every year
−18%
Year-on-year drop in mass IT services fresher hiring
₹3.5 LPA
Standard fresher IT offer — unchanged since 2014
10,000
Applications per fresher opening at a mid-size company
In monthly terms, ₹3.5 LPA translates to roughly ₹18,000–23,000 in hand. Set that against Bangalore rent and an education-loan EMI, and many freshers are cash-flow negative before they buy groceries. Meanwhile, the fees for the private colleges producing those freshers have climbed to ₹8–15 lakh for a four-year programme.
The machine at the gate
At this volume, no human reads resumes. Companies deploy AI screening systems that parse, rank, and reject thousands of candidates before a recruiter opens a single file. The machine that shrank the entry-level job market is now the gatekeeper for what remains of it.
Hold all four numbers together:
15 lakh graduates · 18% fewer seats · salaries frozen since 2014 · 10,000 applicants per opening.
The conclusion writes itself. The generic engineering degree, purchased at 2026 prices, is one of the worst financial products an Indian family can buy.
Here is the part that changes the story from despair to strategy: the same market that offers ₹3.5 LPA to one graduate offers ₹25–40 LPA to another — from the same graduating year, sometimes the same city.
Generic IT services fresher
Barely moved since 2014
Top 10% freshers
Strong portfolio + internship
AI/ML specialist fresher
GenAI, LLM pipelines, MLOps
Freshers with demonstrated AI/ML capability over classical IT roles
Cloud and DevOps salaries in the 2–4 year bracket — fastest band in the entire IT salary table
Internship-to-hire — the single most reliable door into employment that exists
Graduates with real work experience vs. those without
'100% Placement'
A ₹3.2 LPA mass offer counts as 'placed.' Ask for the median salary and how many offers were above ₹5 LPA. The reluctance to share this number is itself the data point.
Foreign immersion photos
Two-week campus tours bundled into the fee at ₹1.5–2 lakh. An AI screening filter has no field for 'visited a foreign campus' — it ignores it in a fraction of a second.
'AI' in the branch name
Many colleges renamed plain CSE to 'CSE (AI)' without changing faculty or content. Verify the GPU lab exists and the faculty are active practitioners.
Pre-2022 syllabus
The most consequential technology shift happened in 2022–23. If the syllabus hasn't been revised since before it, you're paying 2026 prices for 2019 content.
Tutorial clone as 'major project'
If the final-year project is a Django to-do list or a CRUD app from a YouTube tutorial, the college is not preparing students for AI-screened job applications.
The branch name matters less than what is inside it — but some names carry meaningfully better odds:
Specialisation forces newer syllabi by design. Now offered at IITs, IIITs, NITs, and serious private universities.
Many colleges renamed plain CSE to 'CSE (AI)' without changing faculty or content. Verify who teaches ML and whether a GPU lab exists.
Chip design sits where AI meets the physical world. Semiconductor companies keep running fresher drives even as pure-software hiring stalls. India's semiconductor push is policy-backed.
Hardware cannot be prompted into existence — and that's exactly the point.
A structural talent shortage that AI has not closed — security work demands adversarial judgment machines do not replicate. Hiring stayed active through the years developer hiring shrank.
Look for programmes with real red-team/blue-team labs, not just theory modules.
Hired across banking, retail, healthcare, and telecom — not just tech firms. Cloud/DevOps salaries in the 2–4 year bracket grew 24% YoY, the fastest-moving band in the IT salary table.
The most forgiving entry lane for non-CS branches with strong aptitude.
The value outlier: a legitimate IIT credential at a fraction of private-college cost, structured so a student can stack internships alongside it from year one.
Requires discipline and self-direction — ideal for students who learn by building.
Hiring data is unambiguous: graduates with real work experience get hired at roughly double the rate of those without. The degree gets a candidate past the filter; the proof gets them the job.
Create a GitHub profile. Push real code from day one — not just assignments. Start reading documentation, not just watching tutorials.
Land your first internship. Even a small startup counts. The internship-to-hire conversion rate runs at 40–70% — this is the single most reliable door into employment.
Do a serious internship at a company you'd actually want to work at. This is where most hiring decisions are made, quietly, six months before placement season.
Have a specialisation you can defend. One area where you've built something real, can explain your decisions, and have a live link to show. The degree opens the door. The proof gets you in.
Treat college selection like the six-figure purchase it is, and ask the questions a buyer asks:
Ask for the actual course list of semesters 5–8 and check whether it was revised after 2022.
Ask for last year's placement sheet with the median salary, not the topper's package. If the college shares only the highest offer, assume the median is the reason.
Ask whether internships are mandatory and credit-bearing, and which companies took students last year.
Look up the faculty teaching AI and systems courses — their LinkedIn tells you more than the prospectus. An ML course taught by a professor whose last paper was in 2017 is theory, not practice.
Ask for the college's average CGPA-to-offer ratio — how many students above a 7.5 actually received offers above ₹5 LPA. This one question collapses most placement claims instantly.
One rule that compresses all of this:
Fund the checklist, not the brand. A tier-2 college that passes these questions will out-produce a famous one that fails them. The 81%-vs-40% employability variable is built in the corridor between classes, not in the name on the gate.
Polytechnic Diploma (after Class 10)
₹50K–1.5L total
vs. ₹8–15L for private B.Tech
Workshop-heavy. For core engineering — mechanical, electrical, civil — a legitimate complete path. Direct employment at 19 or 20, into work AI is not replacing.
Corporate ATS filters often reject diploma-only candidates for software roles. Exit route: lateral entry into 2nd year B.Tech.
ITI (Industrial Training Institute)
₹10,000–30,000
vs. One-fiftieth the cost of private B.Tech
Teaches a trade honestly — electrician, fitter, welder, machinist, mechatronics — and delivers exactly what it promises: employable skill, apprenticeship pathways, manufacturing and infrastructure hiring.
Not the path for software careers. But for the trades: unmatched value.
The wake-up call is three-directional
Fix the product — curriculum cycles, mandatory internships, median-salary transparency — or keep losing relevance one AI-rejected batch at a time.
Retire the 2014 mental model — 'engineering karo, life set' — and start buying education the way you buy anything else expensive: by inspecting what is inside.
The degree is the entry ticket, not the prize. The internships, the shipped code, the GitHub link — that's what clears the filter.
The queue is 10,000 people long. A machine decides who gets in. The only question: are you paying for what gets your child through that filter — or for the photograph at a foreign gate the filter will never see?
It depends entirely on what is inside the degree. A generic engineering graduate from a college with a pre-2022 syllabus and no mandatory internships faces a brutal market at ₹3.5–4.5 LPA. A graduate from the same year with AI/ML skills, two real internships, and a GitHub portfolio can command ₹25–40 LPA at product companies. The degree is now the entry ticket, not the prize.
Mass IT services hiring — the TCS-Infosys-Wipro volume engine — is down roughly 18% year-on-year, while 15 lakh engineering graduates enter the market every year. Excess supply keeps base salaries flat even as specialised AI/ML salaries surge. The standard fresher offer has barely moved since 2014, but adjusted for inflation it represents a real pay cut.
A ₹3.2 LPA mass offer counts as 'placed.' The number that matters — the median salary of last year's batch and how many offers were above ₹5 LPA — is the number colleges are most reluctant to share. That reluctance is itself a data point. Always ask for the median, not the highest package.
For core engineering (mechanical, electrical, civil, manufacturing), a polytechnic diploma is a legitimate complete path at one-tenth the cost of a private B.Tech. For software careers, corporate ATS filters frequently demand a full-time graduate degree and reject diploma-only candidates. The standard clean exit is diploma followed by lateral entry into the second year of a full-time B.Tech.
B.Tech CSE (AI & ML) with a genuine syllabus revision post-2022, ECE/VLSI for semiconductor careers, Cybersecurity for structural talent shortages, and Cloud/DevOps for the widest sector reach. The IIT Madras BS in Data Science is the value outlier. Across all branches, the four-year execution — GitHub from year one, serious internship in year three — matters more than the admission letter.
Find out if your target job is AI-safe
Take the free AI Risk Quiz — get a personalised risk score and the specific skills to build before the filter runs.
Take the Free AI Risk Quiz →