Vol. I · Reference
Frequently asked questions
A short reference for students, faculty, and anyone evaluating Drona for their classroom.
What is Drona?
Drona is an academic grading platform built for classrooms where students write code. It compiles and runs submissions against instructor-authored test cases, returns rubric-aligned feedback in seconds, and gives faculty a single console for assignments, grades, and integrity review. The goal is to make the feedback loop fast enough that students can iterate the same day a concept is taught.
Who is it for?
Drona is designed for university and college instructors teaching programming courses — introductory CS, data structures, algorithms, software engineering — along with their students and grading assistants. Individual professors can run a single course on it, or a department can standardize across sections. If your workflow already involves running student code against test cases, Drona is meant for you.
How does AI grading work? Is it accurate?
Every submission is executed inside an isolated Docker container against the instructor's test suite; results are deterministic and match what you would get running the tests yourself. On top of that, Drona uses language models to draft rubric-level feedback and annotations, which the instructor can accept, edit, or override before a grade is published. The final grade is always the instructor's — AI proposes, faculty dispose.
How does integrity detection work without being invasive?
Drona combines three passive signals: cross-submission code similarity, a fine-tuned CodeBERT classifier that flags AI-generated code, and an optional keystroke replay of the in-browser editor session. Nothing is recorded outside the editor — no screen capture, no camera, no background monitoring of the student's machine. Flags are surfaced as evidence for the instructor to review, never as automated verdicts.
Is student data private?
Student submissions, grades, and editor sessions are scoped to the class they were produced in and visible only to the enrolled student, the course faculty, and any grading assistants that faculty assigns. Submissions are not used to train third-party models, and integrity signals are kept inside the instructor's dashboard. Data export and deletion requests are honored on the instructor's timeline.
Which institutions can use Drona?
Drona is a standalone product — it is not tied to any single university. Any accredited college, university, or bootcamp can register faculty accounts. During early access we approve professors individually so we can keep the platform trustworthy and gather direct feedback, but the platform itself is institution- agnostic and works with any school email domain.
How do I get access as a professor?
Create a faculty account on the registration page. We verify every educator manually — usually within a business day — to confirm you teach at the institution in your email. You'll get a confirmation email once approved, and then you can create your first class and invite students.
Is it free?
Drona is free for instructors and students during early access. We haven't announced pricing for institutional plans yet; when we do, early-access classes will get meaningful runway to decide whether to continue. Student accounts will always remain free.
Who built Drona?
Drona is built by 5PANDAVA, a team of five engineers who were frustrated by how slow the grading loop felt in their own programming classes. The name is borrowed from the legendary guru who trained the Pandavas — a teacher who expected precision and returned it. We're small, we're shipping quickly, and we read every piece of feedback faculty send us.
Still have questions?
We answer every educator personally.
Register a faculty account and send us the details of your course — we review and respond within a business day.