3-Month AI Plan for Final Year Engineering Students (2026)

Three months is enough time to build a credible AI foundation before your campus placement season opens, provided you spend it on applied work rather than certificates.

That framing matters. Campus drives at most private engineering colleges in India run September through December, with the bulk of tech company visits concentrated in October and November. A final year student starting this plan in July has the right window.

Before the month-by-month breakdown, three honest framings: not every 2026 placement role requires AI skills (service-tier roles at TCS, Infosys, and Wipro still screen primarily on aptitude and coding fundamentals). The free curriculum is genuinely as good as anything paid. And two working projects on a public GitHub beat any number of certificates in a placement conversation.

What three months actually gives you

Twelve weeks at 10–12 hours a week is 130–160 hours of deliberate practice. That is enough to go from zero Python to a deployed ML project. It is not enough to become a senior ML engineer or clear the most demanding product-company AI interviews.

The goal is specific: get to “credible AI fresher” before your campus drive opens. That means you can write and run a Python script, explain how a supervised learning model works, point to a live project on GitHub, and discuss what your model’s output means. That profile clears the initial AI screening at most fresher-level roles and adds substance to any placement interview.

According to a statement by TCS CHRO Sudeep Kunnumal at the AI Impact Summit in March 2026, AI-skilled and higher-skill cadres now account for 60% of TCS’s fresher hiring, up from 10–15% three years earlier. The shift is real, and it is moving fast across the IT services tier.

A rough scope table helps set expectations before you begin:

Prep window	Approx. hours	What you can realistically build
30 hours (1 week)	30	Core ML concepts + interview vocabulary
3 months (12 weeks)	130–160	One deployed project + interview prep
6 months (24 weeks)	250–300	Two projects + deeper track specialisation

For the branch-wise breakdown and role-category detail, the 2026 AI roadmap for Indian engineering students covers the full picture. Worth reading alongside this plan.

Month 1: Python and ML foundations (Weeks 1–4)

The first four weeks have one job: get you to the point where Python is not a barrier. Most students overshoot this phase by trying to learn too many tools at once. The minimum-viable scope for Month 1:

• Weeks 1–2: Python fundamentals. Variables, lists, dictionaries, functions, loops, file I/O. Use a free resource such as Google’s Python class, NPTEL’s Python for Data Science, or any free Great Learning module. You don’t need to finish a full course; you need to be able to write a 50-line Python script without looking up basic syntax.
• Weeks 3–4: NumPy, Pandas, and a first ML model. Install scikit-learn, load a public dataset (the Iris or Titanic dataset from Kaggle works fine), and train a decision tree or logistic regression model. The goal is a Jupyter notebook that runs end-to-end, not a polished model.

Stop switching platforms. Pick one free resource per topic and use it until you have written code. Platform-hopping is the most common way students burn their Month 1 budget without producing anything.

Month 2: One end-to-end project (Weeks 5–8)

One project done well beats five half-finished tutorials. The project should satisfy these criteria:

• Uses a publicly available dataset — Kaggle or a government open-data portal such as data.gov.in
• Trains at least two models and compares their performance on a held-out test set
• Includes a README explaining what you built, why, and what the output numbers mean
• Lives on a public GitHub repository before Week 8 ends

The domain should match your branch where possible:

Branch	Project idea	Public dataset source
CSE / IT	Spam email classifier or movie sentiment analysis	Kaggle
ECE	Audio genre classifier or basic image recognition	Kaggle (GTZAN, CIFAR-10)
EEE	Energy demand forecasting	data.gov.in / Kaggle
Mechanical	Predictive maintenance on sensor readings	NASA Bearing Dataset (Kaggle)
AIDS / Data Science	Customer churn predictor	Kaggle Telco dataset

All of these use free, publicly available datasets and run on Google Colab without a GPU.

NASSCOM’s analysis of India’s talent competitiveness in the AI era consistently points to end-to-end deployed projects as the capability gap between graduates who clear AI screening and those who don’t. Recruiters cannot assess theory in a 30-minute interview; they can assess a GitHub repo.

Keep the scope small enough to finish. A clean sentiment classifier trained on a public movie-review dataset and documented properly is more valuable in a placement interview than an abandoned deep-learning project.

Month 3: Interview prep and portfolio polish (Weeks 9–12)

The third month shifts from building to explaining and presenting.

Portfolio polish (Weeks 9–10)

• Clean the GitHub README: add a one-paragraph summary, the dataset source, the model evaluation metric, and a screenshot or output sample.
• Add a requirements.txt so someone else can run your code.
• Record a 2-minute screen recording walking through the project. Upload it as an unlisted YouTube video or a Google Drive link. Paste the link in your resume.

Interview prep (Weeks 11–12)

Most fresher AI interviews at service-tier companies cover: supervised vs. unsupervised learning, overfitting and regularisation, train-test split, precision and recall, and a walkthrough of your project. Practice explaining each concept in plain language without a whiteboard. The README you wrote in Month 2 is a useful starting script for the project-walkthrough portion.

For product-company AI roles, the bar is higher. Expect Python coding questions on data structures alongside ML concept questions. The 30-hour AI learning plan for placement season 2026 covers the targeted pre-interview sprint in detail, condensing the core into a focused week.

The service-tier reality check

A 3-month AI plan runs parallel to your aptitude and coding prep, not instead of it. TCS NQT, Infosys Infy TQ, and Cocubes aptitude tests are the first gate at most service-tier companies. An AI project on your resume does not help you clear a Cocubes aptitude round if you haven’t prepared for it.

The practical split: 10–12 hours a week total, with 2–3 of those hours set aside for aptitude (quantitative, verbal, logical reasoning) throughout all three months. Don’t defer aptitude to October when your placement season opens.

Not sure which AI role to target before you begin Month 1? The AI engineer vs. data scientist vs. ML engineer comparison for Indian freshers in 2026 lays out what each title actually means on a fresher job description and what the interview expects. That breakdown helps calibrate how deep the Month 3 prep needs to go for your specific target.

If your placement window is already closer than three months and you need a condensed version, the 30-hour AI learning plan for placement season 2026 cuts the plan down to a single focused week, covering exactly the concepts and project signals that fresher AI screening rounds test.