Explore Across Campuses

if you are under 30 in tech: learn one thing AI does not do well for you, and keep it. not a tool. a judgement. the ability to tell when a generated answer is confidently wrong is not a skill that improves with usage of the thing generating it. that's the one to keep

cybersecurity has been the only consistently strong line in indian tech hiring through the entire AI shift. if you are starting now and want a field that will exist in ten years with fewer people in it than the postings suggest, it is that one. boring answer, durable answer

junior devs report feeling less secure about their career than CTOs do: 50% vs 20%. that's inverted from every previous technology cycle, where the seniors were the anxious ones. the people at the bottom of the ladder are the anxious ones now. if you know a junior in tech, that's the conversation to have

80% of developers are under some AI usage limit. token quotas, caps, budget restrictions. a colleague asked me to review a file and apologised for the token cost. we have created a profession where using your own job's tool has a running meter on it. that is not sustainable and it is not a complaint, it is an observation

standalone prompt engineering is now 5-8% of AI hiring. it collapsed into a baseline skill inside 12 months. if you built a 2024 portfolio on 'prompt engineering', you built a portfolio on something that is now listed as a basic competency. that is not your fault. the shelf moved

to anyone reading placement data in 2026: separate the median from the mode and separate the mode from the maximum. the maximum is a person. the mode is what most people got. the median is what half got. companies are recruiting on the maximum and students are anxious about the median. the gap between those two things is where the disappointment lives

the roles actually growing in india right now: AI and ML engineering, data engineering, model validation, AI governance, cybersecurity, cloud architecture, and supervising automated workflows. that last one is new and nobody teaches it. if you are picking a specialisation this year, that list is what the postings say

if you are reading this and you are in your final year: nobody hiring in 2027 is going to ask what your cgpa was. they are going to ask what you have shipped that someone uses. build one thing that has a user who is not you. that is the entire difference between two resumes that look identical

if your employer has announced an 'AI transformation', ask two questions in writing: which roles are being reduced, and what is the intake plan for the next two quarters. the answer to the second one is the real one. if intake is unchanged, nothing is being reduced. if it is zero, something is

for anyone in a GCC support or operations role right now: the risk is not a layoff notice, it is not being replaced when a colleague leaves. that's how it goes. nobody decides to remove your team. people just slowly stop being replenished. watch the intake number, not the attrition number

if your college still runs a single 'placement season' with 3 mass recruiters and a 4.5 LPA average, that is not your placement season. that is a formality with an average attached. do the arithmetic yourself: how many got an offer that was not one of those 3, and at what number

if you are in first year: go to the mess in week one, go to the fest in week one, talk to the person next to you in class in week one. you will not do any of this. i did none of it. i met the people i actually still talk to in the third week, by accident, in a canteen queue. that queue is the only reason this post is not about loneliness.

the honest version of 'AI will take your job' is more specific and more useful: the tasks your job was 80% of are being absorbed, and the 20% is where you need to be. most people can name the 80% in a sentence and have never once written down the 20%

if you are a 2026 graduate in tech: the first rung is going to be harder than it was for the batch above you and easier for the batch below. plan for 12 months of low pay and learn something that compounds while you wait. the wait is the market's, not yours

the cheapest insurance in tech right now is a domain nobody has automated the paperwork for. healthcare admin, agri supply chain, insurance claims, education admin, logistics compliance. these domains have documents, not code. a person who can automate the documents in their own industry is a very short list of people in india

the companies hiring in 2026 are not hiring for what you learned. they are hiring for what you can be made to do by february. learn one skill deeply enough to be useful in 8 weeks, and one broadly enough to be trusted with the rest. that combination is rarer than a 10-CGPA student who has only ever done the syllabus

a recruiter told a room that he doesn't look at github activity as a hiring signal. he'd look at PRs. 'there are other things that spike activity without real work.' he's describing exactly the incentive we created when we made github a portfolio. the fix is shipped work with a reason, not volume

if you are 23-28, in tech, and doing well: the worst mistake available to you is to optimise for the current role. the current role has a shelf life you can look up. spend the surplus effort on something that survives your next job, your next company, and the possibility that the industry you are in changes shape twice in your career

the premium is going to people who combine tech with business judgment. a CTO who can read a requirement, price it, and know which part of it should not be built with an agent. that skill is not taught anywhere. you either get it from a job that gives you real consequences or you do not get it