Work on the gap between what an institution knows and what it does.
We are a small, independent team in New Delhi working on a problem that affects millions of students who will never learn our names. If you want your work to be checkable by a sceptic, this is a good place to be.
Ten people. Five hundred institutions.
That ratio is the appeal and the cost of a team this size, and we would rather you know both before you apply. There is no layer of management between you and the decision.
You own a slice, end to end
No hands-off product managers. You make the call, you ship it, and you sit in the room when an institution asks why it behaves that way.
Meetings are the exception
Decisions get written where anyone can find them later. One short sync a day; almost everything else happens in text, on your clock.
You publish what you find
No internal review can suppress a result for being commercially inconvenient. Two of our six papers contain findings that are bad for us.
What we are hiring for.
We hire slowly. If none of these fit but you think you belong here, write to team@eduplatter.com and say why.
Senior Data Engineer
Integration & warehouse. Own the integration layer: connectors to SIS, LMS, CRM and attendance systems, the warehouse model, and the reconciliation process where an institution’s own reporting is the source of truth. You have finished an integration that other people said was impossible.
Senior Data Engineer
What you’d do
- Build and maintain connectors for the systems Indian institutions actually run
- Own the warehouse model and the versioned data dictionary tooling
- Run reconciliation with institutional IR offices until the numbers agree
- Keep nightly refresh reliable across dozens of tenants
What we’re looking for
- Several years of production data engineering, ideally in a multi-tenant setting
- Comfort reading undocumented vendor schemas and asking better questions
- Patience for the political part of agreeing a definition
Applied Scientist, Student Success
Modelling & fairness. Fit and validate the institution-specific models, run the subgroup fairness reporting, and design the matched-comparison efficacy studies. You care more about whether a result holds than whether it is impressive.
Applied Scientist, Student Success
What you’d do
- Fit persistence, completion and course-risk models per institution
- Own the fairness reporting that blocks deployment on a subgroup gap
- Design and run termly efficacy studies with honest intervals
- Publish the method, including the studies that find nothing
What we’re looking for
- Applied statistics or ML with real causal inference experience
- Familiarity with education data and its particular messiness
- Willingness to tell a client their flagship programme does not work
Implementation Lead
Deployment & partnership. Run deployments end to end: discovery, integration coordination, definition workshops, advisor training and the first efficacy cycle. This is the role that decides whether a project succeeds.
Implementation Lead
What you’d do
- Lead discovery and the data dictionary workshop with registrars and deans
- Coordinate integration between our engineers and institutional IT
- Train advisors on the queue and the closure discipline
- Own the first efficacy study conversation, including null results
What we’re looking for
- Experience deploying software inside universities, or working in one
- Enough technical depth to challenge an engineer’s estimate
- Comfort saying an institution is not ready yet
Product Designer
Product & research. Own the design practice: advisor workflows, student-facing views, the research pages and the brand. The hardest problem here is making a probabilistic prediction feel like useful information rather than a verdict.
Product Designer
What you’d do
- Design the advisor queue and the student view, which have opposite requirements
- Make uncertainty legible without making it ignorable
- Own the design system across product and marketing
- Prototype in code, or close enough for an engineer to read
What we’re looking for
- A portfolio with at least one dense, data-heavy product shipped end to end
- Interest in how people misread probability
- An opinion about what should not be on screen