90.4%.
That’s the share of Batch 2026 that had already interned at real companies before their placement data was even published.
Not after graduation. Before.
Prospectus p.14. Same document, same page, same sourcing as the placement rate that every family asks about first.
Here’s why I think this number matters more than the rate.
What “real company” means in this context
There’s a version of an internship that doesn’t count for much from a hiring perspective.
A three-week shadow programme at a firm a family connection arranged. A summer sitting in meetings with no scope and no deliverable. A certificate attached to watching recorded modules on a company LMS.
Those produce students who can say they interned. They don’t produce students who’ve shipped anything a real team depended on.
The companies in Batch 2026’s internship record are different.
Morgan Stanley. PhonePe. Thoughtworks. Lowe’s. Yellow.ai. Tata 1mg. Maersk. 7-Eleven. Rupeek. Medable. Clari. Raksul. More than 40 recruitment partners spanning nine sectors in total.
These companies have standards. Morgan Stanley’s India engineering team handles real financial infrastructure. PhonePe processes payments for hundreds of millions of users. Thoughtworks is among the more selective engineering employers in India. They don’t put students on real work out of charity. They do it because a student at that point in the programme can contribute something.
When 90.4% of a batch has cleared those screens, the number’s telling you something structural about how the programme was designed.
When the internship happens
Work-integrated learning at Kalvium starts from Semester 3. That’s the first semester of Year 2.
Students choose one of four tracks at that point: Simulated Work, Internship with tech-first companies, Open Source Projects contributing to global codebases, or the Build track for students developing their own products. What Semester 3 at Kalvium actually involves covers each of those tracks and what the shift in structure means from the inside.
The internship track is the one that produces the named placements above. The other three tracks are equally real and demanding. Students on Simulated Work run in-classroom engineering sprints structured to replicate a professional team environment, with briefs and delivery cadences, not exercises with a single correct answer. Students on Open Source are sending pull requests to global codebases, reviewed by maintainers who have no incentive to be encouraging.
The 90.4% is specifically the share of the batch that interned at real companies. Simulated Work, Open Source, and Build are not company internships. Some students in those tracks may also have interned externally. What the figure tells you is that across the full batch, nine in ten students had cleared a real company’s hiring screen and done real work there.
What this means on day one of the first job
I’ve spent a long time thinking about what happens in the first thirty days of a fresh graduate’s first real role.
Most students from traditional programmes arrive on day one having done two types of work: coursework assessed by exams and, if they were lucky, a project or two in final year. They know how to study for a test. They’ve never had a deadline set by someone who doesn’t care about their grade.
They don’t know what happens when code they wrote breaks something in production.
They don’t know the difference between working on a feature in isolation and working in a codebase maintained by eight other people.
They don’t know how professional feedback lands, which is different from teacher feedback in every important way.
The first month of a tech role for a student like that is usually a slow, expensive process of learning those things while the company absorbs the cost.
When a student has already interned at PhonePe or Thoughtworks before their full-time hiring loop begins, they’ve already made those mistakes. They’ve already seen code go to production. They’ve already discovered, in real time with real consequences, what “it works on my machine” actually costs a team when it doesn’t work anywhere else.
That changes what day one looks like. Concretely.
The student can read a codebase they didn’t write because they’ve done it before. They can make an architectural decision without waiting to be told the right answer because they’ve already had to make one under real constraints and live with the result.
When Rajesh and Venkat previously co-founded FACE Prep, we trained students from 2,000 or more institutions for this specific gap. We knew exactly what skills were missing by the time a student reached placement season. We built intensive training to close them in the three months before interviews began. It worked, some of the time. But three months of coaching can’t fully close a four-year doing gap. It was always a patch on a structural problem.
The 90.4% is what happens when the structure addresses the problem from Year 1, not Year 4.
What the rate says about programme design
This number didn’t happen by accident.
Companies don’t extend real internships to students who show up unprepared. The students who interned at Morgan Stanley and Medable had already spent four semesters in a curriculum that starts full-stack development in Semester 1, runs daily coding practice six days a week through the DOJO system, and moves into real company work from Semester 3.
By the time a company is evaluating a Kalvium Year-2 student for an internship screen, that student has a codebase to show. Not a final-year capstone assembled in the last semester, a progression of real builds that started in Year 1 and deepened through Year 2.
That’s what most internship screens actually test for.
Can you show me something you built and walk me through every decision you made? Students who’ve done real work for real teams can answer that question honestly. Students who’ve spent four years in lecture-then-exam cycles are constructing an answer on the spot. The difference shows in the first twenty minutes of any technical conversation.
Where Kalvium CSE graduates actually end up working covers the sector spread from Batch 2026. Nine sectors. No single one above 20%. That spread is partly a product of this same design: students who have already worked in real environments can go in multiple directions, not just the lane their programme was optimised for.
What this does not guarantee
I want to be careful here.
90.4% is a cohort figure from Batch 2026. It tells you what one batch produced across its work-integration tracks in a specific set of hiring conditions. It doesn’t predict what any individual student will experience.
The internship track specifically requires a real company to decide a student’s worth putting on real work. Getting through that screen requires technical readiness that’s built through two years of consistent work: daily coding practice, building things that run, fixing things that break in production during Simulated Work sprints. Students who treat the programme as something that happens to them, rather than something they’re doing, don’t build that record.
The 90.4% is what the programme produced for students who showed up and worked. It’s not a default outcome.
The 82.40% placement rate and median Rs 16.5 LPA, as of March 2026, reflect where those students landed in the full-time market. Manik’s post on how to read Kalvium’s placement numbers honestly covers the methodology behind those figures.
The three numbers belong together: placement rate, sector spread, and pre-graduation internship rate. None of them tells the full story alone.
What the 90.4% adds is the piece that’s hardest to manufacture: students who’ve already been professionals before they became employees.
Rajesh is a co-founder of Kalvium. Rajesh and Venkat previously co-founded FACE Prep, where they worked with 2,000 or more institutions across India. He writes about what it takes to produce engineers who can do real work, and what the data from real batches actually tells you.