Technology at Avanti Fellows

Avanti Fellows helps low-income students qualify into engineering, medicine and other professional colleges. Most of that work happens in classrooms, through teachers, in partnership with government school systems. Technology is how we reach the students we cannot reach in person, and how a small team runs a programme spread across 43 centres in 18 states and union territories.

We build almost all of it ourselves, with a small engineering team, and the code behind what students use is public on GitHub. We have made a deliberate choice to build many small tools rather than one large application. That has a cost - it is a lot of systems for an organisation of our size - but it has let us move quickly and shape each tool around the people who use it.

Our technology falls into three groups: what students use, what our teachers and programme teams run programmes on, and the tools that run the organisation itself.

What students use

• Gurukul, our free online learning platform. ~260,000 students used it in 2025-26. It carries study material, live classes, practice tests and each student’s own performance reports.
• Tests. Our online tests run on our own test engine, so results flow straight back to the student and their teacher instead of being entered by hand.
• Futures, a college predictor, scholarship finder and CV builder. These are open to any student in India, not only Avanti’s, and need no sign-in.
• The scholarship portal, through which students apply for the TATA AIG - Avanti Fellows scholarship and our team reviews and shortlists them in the same place.

What our teachers and programme teams use

• The LMS, the main application for teachers and centre staff. In order to run a centre well a teacher has to mark attendance, track where each batch is in the curriculum, collect student documents and act on test results. These used to sit in separate spreadsheets and tools; the LMS puts them in one place.
• Content authoring. Our curriculum team writes chapters, questions and tests in our own content system, and programme staff schedule tests and live classes in a separate tool. A scheduled test is ready for students without anyone from the tech team getting involved.
• Field campaign tools, for example the app staff use during school outreach campaigns to see their assigned schools and file visit reports.

Underneath these sits a shared core: one record of students, schools, batches and content, and a single sign-in for students. Every programme uses the same core in a different combination, which is why a new state or school system can usually be onboarded without building anything new.

The tools that run the organisation

The third group is easy to overlook, but it is where a lot of our recent effort has gone. As the organisation has grown, much of the work of running it - fundraising, finance, reporting to funders - has needed systems of its own.

• Data. Everything our students do on our platforms, along with public exam results and government datasets, lands in one data warehouse. That lets us answer questions like how many of our students qualify for the JEE, school by school, and compare them fairly with students outside our programmes.
• The data connector. Staff can ask questions of that warehouse in plain English from the AI assistant they already use, and pull the answer straight into a board note or a grant report. It is read-only, every query is tied to a named person, and access ends the moment someone leaves the organisation.
• Fundraising. Our donor CRM assembles itself from the fundraising team’s email and conversations, using AI to keep each donor’s timeline and next steps current, so nobody has to update it by hand.
• Finance, HR and selection. Smaller tools handle grant budgets, the annual feedback cycle and the selection of students into our centres.

How we use AI

AI shows up in three places, and they are at very different stages.

The first is in how we build. A small team maintains a large number of systems, and AI-assisted development is a big part of why that is possible.

The second is in our internal tools, described above. These are working today.

The third is with students, and here we are early. Students who use Gurukul actively do much better, but when we looked closely that looks more like a sign of a serious student than an effect of the platform. So the impact of the online programme itself is still unproven. We are piloting AI-written mentorship reports that would help a mentor understand where each student is struggling, and testing AI-generated worked solutions for our test questions, which our curriculum team checks before anything reaches a student. In both cases a person reviews what the AI produces before it is used.

How we look after student data

Most of our students are under 18, so we are careful with their data. Access to student data is limited to the staff who need it, AI tools can read it but not change it, and access is tied to a person’s organisation account so it ends when they leave. Public education data we have assembled is shared with partner organisations through a separate service that cannot see any student records.

A more detailed, technical description of every system and how they connect is in Technology stack & data flows.