6 million students.
That’s the scale of what Venkat and I saw at FACE Prep, working with 2,000+ institutions across India before we left to start Kalvium.
I don’t throw that number around lightly. I throw it around because it’s the source of a pattern we couldn’t stop seeing. And once that pattern became undeniable, it changed everything about how we designed this programme.
What FACE Prep actually was
Venkat and I previously co-founded FACE Prep, one of India’s largest campus recruitment training companies.
Our model was simple: colleges brought us in during the final semester, three to six months before graduation, to help students clear corporate hiring tests. We ran aptitude preparation, mock interviews, group discussion coaching, resume workshops. We were good at it.
And we sat at the very end of the B.Tech pipeline, watching what four years had produced.
That position taught me something that years of reading research about education couldn’t. When you’re at the front of the pipeline, you see what gets taught. When you’re at the end, you see what lands. The gap between those two views is where all the real information is.
The thing we kept seeing
Here’s what the data kept showing.
Students knew things. They’d covered the right curriculum. Data structures, algorithms, object-oriented programming, databases, operating systems. They could pass multiple-choice screens. Many of them had decent CGPA scores. Knowledge wasn’t the differentiator.
What they couldn’t do was the job.
They couldn’t write code that a colleague could read and extend. They couldn’t open a codebase they hadn’t written and make sense of it. They couldn’t explain what they’d built without rehearsed, vague answers. They couldn’t work inside a team where the outcome depended on their specific contribution. They struggled to recover when an interviewer pushed back hard.
None of that is a knowledge gap. All of it is a doing gap.
Here’s the thing: knowing what a binary tree is and building a system that uses one effectively are completely different skills. You can study the first in an afternoon. The second only forms through repetition, feedback, and real stakes. And most engineering colleges were producing students who had done the first, and almost none of the second.
We saw this across institutions of every kind. It wasn’t about the college’s ranking or location or resources. It was structural. The system was designed to produce graduates who knew things, not graduates who could do things. And no amount of last-semester training was going to fix that at the root.
Three patterns that repeated without exception
We ran enough batches to stop being surprised. But three patterns came back so consistently that I still think about them whenever we’re designing anything at Kalvium.
The communication filter came first, always.
Round one of most corporate hiring loops involves some form of explanation: explain this piece of code, describe how you’d approach this problem, walk me through something you built. The filter isn’t purely technical. It’s whether you can make someone else understand what you know.
Students who couldn’t explain what they’d built in round one almost never made it to round two. And most of them couldn’t explain it cleanly. Not because they hadn’t learned the material. But because explaining your own work to someone who wasn’t there is a completely different skill from learning it yourself. You can’t just know the concept. You have to have put it into words enough times that the words come out clearly under pressure.
No curriculum was teaching both skills in the same setting. The idea that explaining your work is a communication skill, not a technical skill, meant it didn’t get built at all.
The portfolio question separated everyone.
Round two almost always included some version of: walk me through something you built. Not a textbook problem. Not a hypothetical scenario. A real thing you made, every decision in it, what broke, and what you’d do differently.
Students who had built something real, even a small project with a narrow scope, outperformed students who had only studied. The differentiator wasn’t complexity or ambition. It was depth of ownership. Can you explain why you chose this approach over another? What actually broke during testing? What would you change if you rebuilt it today?
Most students had nothing like that. They’d spent four years studying how to build things. They hadn’t built anything you could point to.
Recovery under pressure was always visible.
The third pattern was harder to articulate until we’d seen it enough times. Students who had navigated something genuinely hard before their hiring loop, a difficult semester, a project that failed, a setback they had to work through, carried themselves differently in the interview room. Not dramatically. Not in any way you could put on a rubric. Just a steadiness that students who had only ever coasted didn’t have.
Employability isn’t purely technical capability. It’s also the confidence that comes from having built capability through real difficulty. The two things reinforce each other. And you can’t rehearse the second without having done the first.
What three months couldn’t fix
Our training helped. It moved pass rates on aptitude tests. It improved average interview performance. We were solving real problems.
But three months had a hard ceiling.
You cannot close a four-year gap in three months.
Every improvement we made at FACE Prep was limited by what had or hadn’t happened in the years before we arrived. We could sharpen communication skills. We couldn’t build the history of explaining things that makes communication under pressure feel natural. We could run mock interviews. We couldn’t manufacture the project story for a student who had nothing to walk through.
The deeper problem was systemic. I’ve written at length about what I think of as the operating-system problem in Indian engineering education: the curriculum isn’t the core issue, the administrative architecture around it is. Colleges tracked attendance, not capability. They measured exam scores, not production. They awarded degrees, not demonstrated doing. The students who came to us weren’t failing because they were lazy or incapable. They were products of a system that had never asked them to produce anything real.
That system ran across the 2,000+ institutions we worked with. It wasn’t an outlier at a few low-ranked colleges. It was the norm, everywhere.
What we decided to build instead
After years of watching this from the end of the pipeline, the question changed.
It stopped being “how do we help students clear hiring tests in three months” and became “what would the four years look like if they were designed to close the doing gap, not just the knowing gap?”
That became Kalvium. And three structural choices drove everything.
Production from Year 1. Not from Year 3 when “major projects” begin. Not from Year 4 when the capstone starts. From Year 1, semester 1. Students at Kalvium ship working software before the end of their first semester. The DOJO system runs daily coding practice six days a week, with belt tests that are pass or fail, no coasting on participation. You earn belts in Java, Python, JavaScript, C++ or you earn them again. There’s no middle result. Arvind’s analysis of why most engineering programmes don’t produce engineers lays out the learning science behind why this volume and feedback structure works. The short version: expertise in complex domains doesn’t form through comprehension. It forms through accumulated hours of practice with real feedback.
Squads, not solo. Kalvium students are placed in numbered squads and they work within those squads across their four years. Squad 49. Squad 56. Squad 85. The project either ships or it doesn’t, and your name is on both outcomes. You can’t hide behind a classmate’s effort, and you can’t shoulder the whole thing alone. The communication skills, the interdependence skills, the recovery skills that showed up in our best candidates at FACE Prep: all of them form inside this structure, not in a workshop. You can’t train someone to work well with people through lectures about working well with people.
Real work embedded in the programme from Year 2. From their second year, Kalvium students work on live projects with industry partners. Not internships tagged onto the side. Work built into the architecture of the programme, with academic credit tied to real delivery. By the time a student is in a hiring loop, the six things that actually differentiate CSE freshers in hiring rooms are already in their work history, not their preparation list.
What the numbers say now
The 2026 batch was 82.40% placed as of March 2026, with a median package of ₹16.5 LPA. The floor was ₹15 LPA.
Those numbers need context, and Manik’s post on how to read Kalvium’s placement data honestly walks through what the denominator means, what the sector spread looks like, and the five questions any family should ask any college.
The figure I care about most isn’t the median. It’s the floor.
₹15 LPA is what the programme delivered for the students who weren’t the outliers. Not the Google Summer of Code winner. Not the hackathon sweep. The student with average CGPA who’d still spent four years building real things and could walk an interviewer through every line.
That’s the test of whether the doing gap was actually closed, or whether the headline number just reflects the best twenty percent.
What this data actually means
This is the honest version of the story.
FACE Prep showed us, through 6 million students across 2,000+ institutions, that knowledge was not the constraint. The constraint was doing. Specifically: the absence of four years spent building real things, explaining your work, recovering from real failures, and depending on and being depended on by real people.
End-of-pipeline training could narrow that gap at the edges. It couldn’t reverse a four-year design that had never asked for any of it.
The durable skills that hiring loops actually test for aren’t installed in three months of remediation. They’re built across four years of being put in situations that demand them, with mentors and peers who are watching, and with stakes that are real enough to matter.
FACE Prep was the best answer we could build inside a broken pipeline. Kalvium is the attempt to stop the pipeline from being broken.
The question for any family evaluating a B.Tech programme is simple. Which four years are you paying for? The ones that produce a student who needs three months of remediation before they’re hirable? Or the ones that produce a student who walks into a hiring room already able to walk through what they built?
That gap is entirely a design choice. Every college makes it, whether they acknowledge it or not.
Rajesh is a co-founder of Kalvium. He and Venkat previously co-founded FACE Prep, where they worked with 2,000+ institutions and 6 million students across India. He writes about higher-education reform, what it takes to build engineers instead of graduates, and the long arc of making Indian engineering education worth the four years it asks for.