For ria.

Coverage: 2027 batch · India on-campus hiring · 7.5/11 LPA communication round · 16 LPA root_mind OA and interview · reports from September 2026

About Next Round

Your next round is most likely a Technical OA, but Capgemini is using more than one assessment path this year.

1. Check the words in your invite. The section names tell you more than an old prep video or a package rumor.
2. For the new Exceller path, prepare AI and technical MCQs, debugging, AI-assisted coding, and cognitive tests. This is the strongest match for current 7.5/11 LPA campus schedules.
3. For root_mind or 13/16 LPA, prepare for a more hands-on test. One recent candidate had DSA, prompt writing, bug fixing in an existing project, and feature work.
4. Some drives still use the older pattern. If your invite says Pseudocode, IT Fundamentals, two coding questions, and Gaming, follow that instead.
5. There is no trustworthy public cutoff or package formula yet. Do not plan around claims such as “this many answers guarantees 11 LPA.”

Safest weekend formula: 30% technical and AI MCQs · 25% debugging · 20% AI-assisted coding · 15% one timed DSA problem · 10% cognitive practice.

First, find your assessment track

If the email is vague, prep them all!💪

What to prepare for the new Exceller OA

The points below come from a shared assessment framework and current campus material. They explain the likely structure. Colleges can still receive different versions though.

1. Technical Module

The reported format is 40 MCQs in about 45 minutes.

You get a little over one minute per question. Finish direct definition questions first. Do not lose three minutes on one long code trace.

A few facts worth knowing:

• RAG retrieves useful information before the model answers. It can reduce hallucination, but it does not guarantee a correct answer.
• An AI agent can plan several steps, use tools, check results, and adjust what it does next.
• An API key normally identifies an app or project. An access token normally carries limited authorization.
• A Docker container shares the host kernel. A virtual machine includes a guest operating system.
• WHERE filters rows before grouping. HAVING filters groups after GROUP BY.
• BFS uses a queue. DFS uses a stack or recursion.
• Binary search is O(log n) on sorted data. Hash-table lookup is O(1) on average.
• ACID means Atomicity, Consistency, Isolation, and Durability.

2. Debugging Assessment

The shared format says:

• One question in about 20 minutes.
• C, C++, and Java are listed. Reports about Python still conflict.
• You receive code that already contains a problem. You are not starting from a blank editor.
• The expected flow is Review → Identify → Fix → Validate.
• Trees, graphs, 2D dynamic programming, and other DSA topics are listed as possible areas. That does not mean every candidate gets them.

Look for simple bugs before rewriting the solution:

1. Wrong loop boundary or array index.
2. Missing recursion base case.
3. A visited node marked too late in BFS or DFS.
4. Wrong DP table size or initialization.
5. Wrong comparison, return value, heap direction, or data type.
6. Null access, division by zero, or integer overflow.

A simple 20-minute split: three minutes to read, six to trace, five to fix, and six to test normal and edge cases.

3. AI-Assisted Coding

This round checks how you work with the built-in assistant, not only whether the final code runs.

The framework lists four things:

1. Understanding: Did you read the problem correctly?
2. Prompt quality: Did you give clear instructions?
3. Problem solving: Did you choose a sensible approach?
4. Review and adapt: Did you test and improve the code?

Use a prompt like this:

State the goal, input, output, constraints, edge cases, language, and required function signature. Ask for the approach and complexity before asking for code. Test the result. If a case fails, show the exact input, expected result, and actual result before asking for a small fix.

Do not type only “write the code.” Do not submit code you cannot explain. During the test, use only the assistant provided inside the assessment.

4. Cognitive Assessment

For Motion, plan before moving. For Grid, remember positions in small groups. For Behavioral, answer honestly and stay consistent. There is no secret personality answer key.

If every listed stage runs in one sitting, prepare for roughly 2.5 to 3 hours, including transitions. Your own invite is still the final source for timing.

If your invite says root_mind or 13/16 LPA

Skip this section if your invite does not mention a high-package role, framework choice, bug fixing, or feature development.

One recent 16 LPA root_mind candidate reported:

1. DSA — 25 minutes: one easy-to-medium coding problem. The candidate received a frequency-array task and said some other slots leaned toward greedy problems.
2. Prompt Engineering: write a prompt that produces the required output.
3. Bug Fixing — 25 minutes: choose a framework such as React, Spring, or Node.js, then fix a stated bug in an existing project with help from the built-in assistant.
4. Feature Development — 40 minutes: stay in the same project and add a requested feature.

This is one firsthand experience. It is useful for the high-package path, but it should not be treated as the paper for every 4.25–11 LPA candidate.

A very simple two-day plan

If the test is on Monday, do not try to learn everything from zero.

Saturday

1. 90 minutes — AI Literacy: LLMs, hallucination, prompts, RAG, agents, bias, and output checks.
2. 90 minutes — Technical basics: OOP, DBMS, SQL, REST, HTTP, Git, Docker, and common complexities.
3. Two hours — Debugging: fix three broken programs under a 20-minute timer, then review your mistakes.
4. 45 minutes — AI-assisted coding: solve one problem by writing a clear prompt, checking the code, and testing edge cases.
5. 30 minutes — Cognitive: learn the Motion and Grid rules. Familiarity matters more than hours of grinding.

Sunday

1. Take one 40-question / 45-minute mixed mock.
2. Review every wrong answer before taking another mock.
3. Do one tree or graph debugging task and one DP debugging task.
4. Solve one easy-to-medium array, hashing, or greedy problem in 25 minutes.
5. Review your strongest allowed language. Do not depend on Python until the test screen confirms it.
6. Stop early enough to sleep properly. The cognitive section punishes tired mistakes.

Before the test

Review only the short list: RAG, agents, prompt structure, OOP, ACID, joins, HTTP methods, Git, Docker, BFS/DFS, and common time complexities. Check your charger, internet, camera, microphone, ID, and test-browser instructions. Do not start a new topic that morning.

What three candidates actually reported

01: Communication round for a 7.5/11 LPA drive

Report date: September 25, 2026 · Process: 2027 campus hiring · Result: Not disclosed

The candidate took the test online from home. It had six parts:

1. Short statement-and-conclusion questions.
2. Two reading passages with about five questions each.
3. Listening tasks followed by MCQs.
4. Two image-comparison speaking tasks, with about 60–90 seconds to speak.
5. Ten sentence-correction questions.
6. Two writing tasks: an email and a project update for stakeholders.

The test used one overall timer. The candidate said it was manageable, but time could disappear quickly. A Safe Exam Browser was used, and the camera, microphone, and screen were recorded.

02: root_mind OA with React

Report date: September 25, 2026 · Role: root_mind · Package: 16 LPA · Stage: OA

The candidate completed one DSA task, a prompt-writing section, an AI-assisted bug fix in a React project, and a new feature in the same project. Java and C++ were available for the DSA section. The candidate was not sure whether Python was offered.

03: Selected root_mind interview candidate

Report date: September 24, 2026 · Role: root_mind · Package: 16 LPA · Result: Selected

The candidate said the interview focused on Computer Networks, Operating Systems, DBMS, and projects. They also said HLD appeared elsewhere in the interview pool, but did not give an exact HLD prompt or say it was asked in every interview.

How sure are we (who are we? we are Jane and ria 👏?

Overall confidence: about 75–85%. The broad direction is solid. Exact details can still change by college, role, and test slot.

What Jane and ria still do not know

• The exact cutoff for each stage.
• How scores map to 7.5 or 11 LPA.
• Whether Python appears in every test location.
• Which DSA or debugging topic each slot will receive.
• Whether every college gets the same order and timing.

Final checklist for ria only

☐ Read your invite and identify the assessment track.

☐ Confirm the allowed coding languages.

☐ Review AI Literacy and the core technical list.

☐ Fix at least two unfamiliar programs under a 20-minute timer.

☐ Complete one guided AI-coding practice without blindly accepting the output.

☐ Learn the Motion and Grid rules.

☐ Prepare your device, ID, test browser, camera, microphone, and internet.

☐ Sleep properly before the test.

Sources: from anywhere useful interview info could be found~~
Good night~
Wait... I still have Airbnb to take care of.
Can’t say goodbye just yet.