Runs the code
Executed in a sandbox against your test cases — hidden ones too.
Less grading.
More teaching.
GradeME AI runs your students' code, reads their written answers, and scores quizzes automatically — while you keep the final say on every mark.
Reads the answer
Scored on meaning against your model answer, not just keywords.
Scores the quiz
Instant marks with per-question analytics and answer explanations.
Powered by a serious grading stack
Judge0Sentence-BERTGeminiMongoDBNext.jsPython
The grading engine
Three grading brains, one gradebook.
Every question type is graded the way it's actually written — not forced through one generic checker.
Code grading
Code that actually runs.
Students' programs execute in a sandbox against your test cases — including hidden ones — and are marked on real output, not guesswork.
- Python, JavaScript, Java, C# and C++
- Function- and class-level checks, not just stdin/stdout
- Per-test breakdown students can learn from
1 function isEven(n) {
2 return n % 2 === 0;
3 }
4 // graded against 5 hidden cases
isEven(4) → true
isEven(7) → false
isEven(0) → true
All tests passing10 / 10
0%similarity
Keywords matched against your model answer
1NF2NFfunctional dependencyredundancycandidate key
Written answers
It reads for meaning.
Theory answers are scored on how closely they match your model answer in meaning — plus the key terms you expect — so a right answer in different words still gets the marks.
- Semantic similarity via Sentence-BERT
- Keyword coverage you define in the rubric
- Low-confidence answers routed to you
Quizzes & exam mode
Quizzes that hold up.
Generate a 10–50 question MCQ quiz from a topic or your notes, run it in a locked exam mode, and see exactly which questions tripped the class up.
- Tab-switch detection + timed auto-submit
- Instant scoring with answer explanations
- Per-question analytics to spot weak spots
Hardest questionQ3 · Normalization
Inside a grade
See exactly why every mark was given.
No black box. Every submission leaves a trail — watch one travel the grading pipeline from upload to gradebook.
-
Submission receivedsumEven.py · 2.1 KB · attempt 1 of 114:02:11
-
Code executed — 4/4 tests passedJudge0 sandbox · 0.8s · incl. 1 hidden case14:02:12
-
Comment quality checked — 87% matchSentence-BERT vs. your model answer14:02:13
-
Confidence check passed96% confident — no manual review needed14:02:13
-
9.4
Posted to the gradebook9.4 / 10 · you can override it with one click14:02:14
And a lot more
Everything a course needs.
Lab manuals, parsed
Upload a PDF and it splits into tasks with the grading data drafted.
AI question generation
Describe a topic and get a full quiz in seconds.
Manual review queue
Anything low-confidence lands in one tidy queue for your call.
Cohort analytics
Class and per-student performance at a glance.
One-click regrade
Override any score, or re-run a failed grading in seconds.
Admin & audit
Roles, deletion requests and a full audit trail.
How it works
From assignment to graded, in three steps.
You stay in the loop the whole way. Anything the AI can't grade with confidence is routed to you.
01
Create
Create or upload
Write an assignment or upload a lab manual — the AI drafts tasks, test cases, model answers and quizzes for your review.
02
Submit
Students submit
Students take quizzes in a locked exam mode, or upload code and written answers before your deadline.
03
Graded
Graded instantly
Results land the moment they submit, with analytics and a one-click override always in your hands.
Loved on campus
Built for the people who grade.
"A lab that used to eat my whole Sunday is graded before I finish my coffee. I just skim the review queue and I'm done."
Sana R.Lecturer · Programming Fundamentals
"The per-question quiz breakdown told me exactly which topic to reteach. My section's scores jumped the very next week."
Prof. Naveed A.HOD · Computer Science
"Getting my code result in seconds — with the exact failing test — beats waiting two weeks for a red pen."
Hamza K.BSCS student · Semester 4
"I uploaded our old lab manual PDF and it came back as sixteen ready-to-publish tasks. I expected to fix everything by hand — I fixed two."
Bilal A.Instructor · Database Systems
"Exam-mode quizzes ended the tab-switching problem in my section. Full stop."
Mahnoor T.Lab Engineer · Web Engineering
"As a TA I used to spend evenings running everyone's code by hand. Now I only open the submissions the AI isn't sure about — maybe three per lab."
Usman J.Teaching Assistant · DSA
"The similarity score plus keyword view means I can defend every theory mark when a student asks. That transparency matters."
Dr. Amber S.Assistant Professor · OS
"Instant feedback changed how I study. I fix my code the same evening instead of forgetting the lab ever existed."
Areeba N.BSCS student · Semester 3
"We piloted it in two lab courses; the audit log and deletion requests made it easy to say yes as an admin."
Kamran H.Department Admin
See it in action
Two minutes, submission to grade.
Watch a real assignment move through the platform, end to end.
GradeME AI — walkthrough.mp41:58
2-min walkthrough · 1080p
Created, submitted, and graded — one continuous take.
Questions
Good to know.
Does the AI grade everything on its own?
No — it does the heavy lifting and you keep the final say. Anything it can't grade with confidence goes to a manual review queue, and you can override any score with one click.
Which programming languages can it run?
Python, JavaScript, Java, C# and C++ execute in a sandbox against your test cases, including function- and class-level checks — not just print-and-compare.
How does it grade written answers?
Theory answers are scored on semantic similarity to your model answer (via Sentence-BERT) plus the keyword coverage you set in the rubric, so a correct answer phrased differently still earns the marks.
Can students cheat on the quizzes?
Quizzes run in a locked exam mode with tab-switch detection, a countdown that keeps running in the background, and auto-submit — plus one attempt per student.
Do I have to write every question myself?
Only if you want to. Describe a topic or upload your notes and GradeME AI drafts test cases, model answers, or a 10–50 question quiz for you to review and edit.
What can students actually submit?
Source files for code tasks (.py, .js, .java, .cs, .cpp and more), and documents (PDF, DOCX or plain text) for written work. Each task validates the file type before it's accepted.
What happens if the AI gets a grade wrong?
You can override any score in one click, and failed code runs can be regraded instantly. Low-confidence grades never publish silently — they wait for you in the review queue.
Do students need to install anything?
No. Everything — submitting, quizzes, results and feedback — runs in the browser. Students sign up with their roll number, batch and section, and they're in.
Grade your first assignment today.
Set it up in minutes. Your students get instant feedback — you get your time back.
Works in your browser · Nothing to install.