Every placement season, the same thing happens in CS departments. Students who passed every lab practical sit their first online coding round and freeze. On paper they could write a program from memory. In the round, the code has to run, and it is checked against inputs they never see.
The fix is practice in the same format, once a month, starting well before placements. What stops most departments is the setting. Finding questions, balancing difficulty and checking every one takes an evening, and nobody has an evening a month to give.
That is the part AI now does. Here is a mock round built end to end on our demo college, with every step shown as it actually came out.
Step 1: Ask Teemly AI for the round
In the question library, we pressed Build with AI and typed one sentence:
A few seconds later, Teemly AI returned a draft paper of seven questions, all taken from the department's own question bank:
| # | Type | Question |
|---|---|---|
| 1 | MCQ | What is the index of the first element of an array in C and Python? |
| 2 | MCQ | A stack follows which order? |
| 3 | MCQ | Time to access the i-th element of an array? |
| 4 | MCQ | In-order traversal of a BST gives keys in? |
| 5 | Coding, medium | Write a function reverse_words(s) that reverses the order of words in a sentence. |
| 6 | MCQ | Deleting a node given only a pointer to it (not the last) takes? |
| 7 | Coding, hard | Write a function second_largest(nums) that returns the second largest distinct value in a list. |
The draft exactly as Teemly AI built it on our demo college's question bank.
It also left a note, which is worth reading every time:
That note is the AI doing its job properly. It did not quietly pass off JavaScript-tagged questions as Python ones. It told us, so we could decide. If the bank had been short of a topic, it would have said that too, instead of making a question up. Teemly AI only ever picks questions that already exist in your bank. It never writes them.
Step 2: Spend five minutes checking the picks
Nothing reaches students until you review the draft and send it, and this is where the review earns its five minutes. Open each coding question and read its test cases, because in a coding round the test cases are the marking scheme.
In our demo bank, the "reverse the words" question had only a placeholder case, so we wrote five before sending it:
| Case | Input | Expected | Seen by student? |
|---|---|---|---|
| 1 | hello world | world hello | Yes |
| 2 | one | one | Yes |
| 3 | I love Python | Python love I | Yes |
| 4 | a b c d e | e d c b a | No |
| 5 | Placement season starts in July | July in starts season Placement | No |
Hold on to case 4. It turns out to be a mistake, and the grading below shows why.
See this in your own lab. Teemly sets coding tests from your question bank, marks every program against test cases, and explains the results in plain English. We set it up free.
Book a free demoStep 3: A format to copy
Campus coding rounds vary by company, but most are a short timed paper of a few coding questions, sat online and checked against hidden cases. A mock round should feel the same. This is the shape our sentence asked for:
| Part | What | Suggested time | Marks |
|---|---|---|---|
| Warm-up | 5 MCQs on data structures | 15 min | 10 |
| Coding 1 | One medium question | 30 min | 20 |
| Coding 2 | One hard question | 45 min | 30 |
| Total | 90 min | 60 |
Run it in the lab, not as homework. The test runs in full screen, copy-paste is blocked and switching windows is recorded for the teacher to review. Answers are saved on the computer as students type, so a short internet drop does not cost anyone their work.
Step 4: The marks come from the test cases
Once a student submits, each program runs against all five cases, and the marks are the share of cases passed. These are four answers students commonly write for this question. We ran each one through Teemly's grader, with the question worth 10 marks:
| Answer | Visible passed | Hidden passed | Marks out of 10 | What happened |
|---|---|---|---|---|
| A: reverse the whole string with s[::-1] | 0 of 3 | 1 of 2 | 2 | Reverses the letters, not the words. Still passed case 4. |
| B: split, reverse the list, join | 3 of 3 | 2 of 2 | 10 | - |
| C: build the answer in a loop | 3 of 3 | 2 of 2 | 10 | Leaves a space at the end, which the grader ignores |
| D: reverse each word's letters | 0 of 3 | 0 of 2 | 0 | Words stay in order, letters flip |
Every answer run through the platform's own grader. Marks are the share of the five cases passed.
Answer A is the lesson. Reversing the whole string turns "a b c d e" into "e d c b a", which happens to be the right answer, because every word is one letter long. So a wrong answer collected 2 marks from a hidden case. Case 4 should have used longer words. A weak case does not just miss a mistake. It pays for one.
That is why Step 2 matters even when the AI picked well. The AI chooses the questions. The test cases are still a teacher's judgement, and writing hidden test cases that catch the mistakes students actually make covers how to get them right.
Step 5: Ask Teemly AI what the round showed
The marks tell you who did well. The next question is what to do about it, and you can now ask the reports directly. On our demo college's CSE 2A, a class of 45 with a semester of marks, we asked two questions.
First question: "Which topics should I revise with this class before their first placement-style coding round?"
Teemly AI did not pretend to know. It said the data had no information about what a placement round covers, so it could not say which topics that round would test. Then it gave what the marks could show: the class's topics from weakest to strongest, with Lists & Dictionaries lowest among the busy topics at 46%.
Second question: "Which topics with plenty of answers is this class weakest in, and how many students are below 30% in the weakest one?"
Teemly AI's answer, with the 15 student names shortened here.
Two details in that answer are deliberate:
- It would not count the students. Every number Teemly AI gives is checked against the marks before you see it. A total it worked out itself could not be checked that way, so it listed the students and left the counting to you.
- The names never left Teemly. Before the question went to the AI, every student was replaced by a label like "Student 7". The names were put back on our screen afterwards.
That gives the next step: one revision session on lists and dictionaries for those 15 students, then a short retest sent only to them.
What Teemly AI does, and what it does not
| Teemly AI does | Teemly AI does not |
|---|---|
| Build a draft paper from your question bank in one sentence | Write or change questions |
| Tell you where the bank is short, or where a pick needs checking | Send anything to students without your approval |
| Answer questions about your marks in plain English | Mark programs. The test cases do that. |
| Say so when the marks cannot answer a question | See student names |
The ten minutes, added up
Typing the sentence took under a minute. Reading the draft and the AI's note took two. Checking the coding questions' test cases took about five, and longer the first time a question is used. The marking took no time, and asking what to revise took one more minute. That is what makes a monthly mock round realistic in a department where nobody has a spare evening.
Try it
To see a coding round from the student's side, the two-question sample test at teemly.in/try has hidden test cases and needs no sign-up. If the round is part of a larger practical, how to run a programming lab exam online covers the full plan. And to run this with your own question bank, this is what Teemly's coding assessments are built for.