GATE DA 2027: Data Science and AI Syllabus, Dates and Pattern

Exam dates

6–21 Feb 2027

Three weekends

Conducted by

IIT Madras

Organizing Institute

Registration opens

27 Aug 2026

Closes 27 Sep 2026

Result

19 Mar 2027

Score valid 3 years

GATE Data Science and Artificial Intelligence — paper code DA — is the newest of the major GATE papers, first conducted in 2024. It exists because AI and data science postgraduate programmes had outgrown the CS paper as an admission filter, and it is now the natural route into dedicated AI, ML and data science programmes at the IITs and IISc.

Being a young paper changes how you should prepare for it. There is a far smaller archive of previous year questions than for CS or ME, the examiners are still establishing conventions, and the official syllabus therefore carries more weight relative to past-paper pattern-matching than it does elsewhere. This page covers what IIT Madras has officially confirmed for GATE 2027 and how to plan around that reality.

What changed for GATE 2027

GATE 2027 is not a copy-paste of the previous cycle. IIT Madras revised the syllabus after five years to align it with current academic and industry practice, and restructured part of the paper list. Three changes matter when you plan:

1. 1The syllabus was revised. This is the first revision in five years and it applies across papers. Any study plan, book list or PYQ set built for GATE 2022–2026 may now contain topics that have been dropped, and may be missing topics that were added.
2. 2Robotics and Automation (RA) is a new test paper. It joins the paper list as a standalone paper.
3. 3Textile Engineering and Fibre Science moved. It is now a sectional paper under Engineering Sciences (XE) rather than a standalone paper.

GATE 2027 exam dates and registration schedule

IIT Madras released the GATE 2027 Information Brochure v1.0, released 20 July 2026. The examination runs across three consecutive weekends in February 2027, in two sessions each day — forenoon 9:30 AM to 12:30 PM and afternoon 2:30 PM to 5:30 PM.

The paper-wise and session-wise timetable is published separately by the Organizing Institute closer to the examination, so which of the six days your paper falls on is not something anyone can tell you yet. Treat any site claiming a confirmed paper-wise date today as unverified.

GATE 2027 DA exam pattern and marking scheme

Data Science and Artificial Intelligence (paper code DA) is a three-hour computer-based test carrying 100 marks across 65 questions — 10 General Aptitude questions and 55 subject questions. Questions carry either 1 or 2 marks.

Duration

3 hours

Computer-based test

Questions

65

10 aptitude + 55 subject

Total marks

100

1 or 2 marks each

Language

English

All papers

Note the structural difference from the engineering papers: DA has no separate 13-mark Engineering Mathematics section. Mathematics is not a bolt-on here — probability, statistics and linear algebra sit inside the subject syllabus itself and are examined as core content.

Three question types appear, and they do not carry the same risk. Knowing which is which on screen is worth real marks:

GATE DA 2027 syllabus structure

The authoritative document is the official DA syllabus PDF for GATE 2027 published by IIT Madras. With a young paper and a freshly revised syllabus, this PDF matters more than any secondary summary — including this one.

DA has been organised around the areas below, alongside the General Aptitude section common to every GATE paper. Confirm topic-level detail against the official 2027 PDF.

• Probability and Statistics — counting, probability axioms, conditional probability and Bayes' theorem, random variables and standard distributions, expectation and variance, the central limit theorem, confidence intervals and hypothesis testing.
• Linear Algebra — vector spaces, matrices, systems of linear equations, rank, determinants, eigenvalues and eigenvectors, and matrix decompositions including SVD.
• Calculus and Optimization — functions of a single variable, limits, continuity and differentiability, Taylor series, maxima and minima, and single-variable optimisation.
• Programming, Data Structures and Algorithms — Python programming, arrays, stacks, queues, linked lists, trees, hash tables, graphs; searching and sorting, divide and conquer, greedy methods, dynamic programming, graph algorithms.
• Database Management and Warehousing — the relational model, relational algebra, SQL, integrity constraints, normalisation, file organisation and indexing, and data warehousing and OLAP concepts.
• Machine Learning — supervised learning including regression, classification, trees and ensembles, support vector machines and neural networks; unsupervised learning including clustering and dimensionality reduction; bias-variance, regularisation and model evaluation.
• Artificial Intelligence — search (informed, uninformed, adversarial), logic and knowledge representation, and reasoning under uncertainty.

The full subject → topic → subtopic breakdown we use for practice and progress tracking is on the GATE DA syllabus page.

How to prepare for GATE DA 2027

That single difference is the most common planning error made by candidates switching from a CS-style preparation. In DA, the mathematics is the paper.

Build mathematics first, then machine learning on top of it

1. 1Probability and Statistics first, and thoroughly. It is examinable in its own right and it underwrites most of machine learning — maximum likelihood, Naive Bayes, evaluation metrics and uncertainty all reduce to probability.
2. 2Linear Algebra next. Eigen-decomposition and SVD are directly examinable and are also the mathematics behind PCA and dimensionality reduction.
3. 3Calculus and Optimization, which is where gradient-based learning comes from. Understanding why gradient descent works makes a whole class of ML questions straightforward.
4. 4Programming, Data Structures and Algorithms. Python-based, and a reliably scoring area for anyone with a CS background.
5. 5Machine Learning, once the three mathematical areas above are solid. Attempted earlier, it degenerates into memorising algorithm names.
6. 6Artificial Intelligence and Database Management and Warehousing, both comparatively self-contained and well suited to a later phase.
7. 7General Aptitude daily — 15 marks, and the most predictable section in the paper.

Prepare for conceptual questions, not library recall

GATE DA examines understanding, not tooling. Questions target why an algorithm behaves the way it does — how regularisation trades bias against variance, why a kernel makes a problem separable, what a confusion matrix implies about a classifier — rather than the API of any particular library. Being fluent in scikit-learn is not preparation; being able to derive a decision boundary is.

Adapt to a thinner PYQ archive

Because DA is a recent paper, there are only a few years of previous questions available, and a syllabus revision reduces how much of even that is directly transferable. Practical consequences:

• Work the official syllabus line by line and make sure you can answer a question on each line. With little past-paper signal, the syllabus is your best predictor of scope.
• Solve every available DA past paper carefully rather than quickly — they are a scarce resource and worth more per question than in an older paper.
• Supplement with standard textbook problems in probability, linear algebra and machine learning, which is where much of the underlying question style originates.
• Do not over-fit to the handful of existing papers. A young examination's pattern is not yet stable, so breadth across the syllabus beats narrow pattern-matching.

Manage risk the way the marking scheme demands

• Attempt every MSQ and NAT question — neither carries negative marking.
• Eliminate before guessing on MCQs; the 1/3 and 2/3 penalties make blind guessing negative-expectation.
• In numerical probability and ML questions, watch the rounding instruction — NAT answers are keyed on a virtual keypad and a correctly-derived value in the wrong format scores nothing.

Audit mocks by subtopic

Trace each wrong answer to its specific subtopic and revise exactly that. Tracked against the GATE DA syllabus tree, this keeps a small PYQ archive from becoming a false sense of readiness.

Eligibility, fees and number of papers

The brochure sets a single academic bar: a candidate currently studying in the third or higher year of any undergraduate degree programme, or who has already completed a government-approved degree in Engineering, Technology, Architecture, Science, Commerce, Arts or Humanities, is eligible. There is no age limit and no cap on the number of attempts.

• Papers: you may appear in one or two test papers. A second paper must be chosen from the allowed two-paper combinations listed in the brochure — the combinations are not free-form.
• Score validity: a GATE 2027 score stays valid for 3 years from the date the results are announced.
• Mode: computer-based test, in English only, conducted at centres across 300+ cities in India. There are no international centres for GATE 2027.
• Two papers: the fee payable is exactly twice the single-paper fee.

Age limit

None

No cap on attempts either

Papers allowed

1 or 2

From the allowed combinations

Score validity

3 years

From the result date

Exam centres

300+ cities

India only, no overseas centres

Registration happens only through the GATE Online Application Processing System (GOAPS), reachable from the official website. Nothing is submitted on paper — IIT Madras explicitly asks candidates not to post printed application forms or documents.

Practising GATE DA previous year questions

Previous year questions are the only fully reliable signal of how DA is actually examined — how deep a topic goes, which standard results get reused, and how a two-mark question is built out of a one-mark idea. Solving a paper untimed and then reading the solution is revision. Solving it under a three-hour clock, then auditing every wrong answer against the topic it came from, is preparation.

Success Tracker keeps a topic-tagged GATE DA PYQ archive with worked solutions, so you can pull every past question on a single subtopic instead of hunting through whole papers. Start from the GATE DA PYQ archive or the full GATE DA syllabus tree.

Frequently asked questions

Who is conducting GATE DA 2027?

IIT Madras is the Organizing Institute for GATE 2027. The exam is conducted jointly by IISc Bengaluru and the IITs on behalf of the National Coordination Board – GATE, Department of Higher Education, Ministry of Education, Government of India. The official website is gate2027.iitm.ac.in.

What is the GATE DA 2027 exam date?

GATE 2027 is scheduled across 6, 7, 13, 14, 20 and 21 February 2027, in forenoon (9:30 AM to 12:30 PM) and afternoon (2:30 PM to 5:30 PM) sessions. The specific day and session for the DA paper is announced closer to the exam and is not yet public.

Does GATE DA have an Engineering Mathematics section?

No. DA is one of the papers where General Aptitude carries 15 marks and the subject syllabus carries the remaining 85, with no separate 13-mark Engineering Mathematics section. Probability, statistics, linear algebra and optimisation are examined as part of the core DA syllabus itself.

Should I choose GATE DA or GATE CS for AI and data science programmes?

It depends on the programmes you are targeting, since each institute specifies which paper it accepts for which programme. DA is designed around data science and AI content, while CS is broader in systems and theory. Check the admission criteria published by the specific departments you want, and note that you may appear in two papers only if the combination is on the official allowed list.

Is there negative marking in GATE DA 2027?

Negative marking applies only to Multiple Choice Questions: 1/3 mark for a wrong 1-mark MCQ and 2/3 mark for a wrong 2-mark MCQ. MSQ and NAT questions carry no negative marking, and no question type awards partial credit.

How long is a GATE 2027 score valid?

A GATE 2027 score remains valid for three years from the date the results are announced. Results are scheduled for 19 March 2027.