Skip to content

Expert: deep learning and AI · Ethics, privacy and careers · topic 61 of 62 · 30 min

Portfolio, projects and interviews

Explanation

Employers want proof that you can turn messy data into useful answers. Build two or three solid projects with real public data, each with a clear question, honest evaluation and a readable write-up on GitHub. Prepare for interviews that test SQL, Python, statistics, ML concepts, a case study and how you communicate.

Why a portfolio beats a list of courses

Thousands of applicants list the same course certificates. A project shows that you can pick a question, find and clean data, analyse or model it sensibly and explain the result. Interviewers often spend the whole conversation on one good project, so a few deep projects beat many shallow ones.

Where to find real data

  • India's Open Government Data Platform (data.gov.in): agriculture, rainfall, education, health and more.
  • RBI's Database on Indian Economy (DBIE) for economic and banking statistics.
  • Kaggle datasets and competitions (check each dataset's licence).
  • Your own data, collected ethically: a survey of your college, prices you track, public transport timings.

What makes a strong project

  1. A real question someone would care about ("Which districts saw the biggest change in monsoon rainfall?"), not "I applied 10 algorithms to the Titanic data".
  2. Honest data work: cleaning steps documented, limitations stated.
  3. A baseline and proper evaluation (module 6), with the right metric explained.
  4. Clear visuals and a short written conclusion: what you found, how sure you are, what to do next.
  5. Reproducible code: a README with how to run it, requirements, and no secrets in the repository.
  6. Bonus: a small deployed demo (a web app or dashboard) for one project.

The interview loop

RoundWhat it testsHow to prepare
SQLJoins, GROUP BY, window functions on business dataDaily SQL practice; the SQL course on this site
Python / codingpandas manipulation; sometimes DSA basicsSolve small data tasks; the DSA course for engineering roles
Statistics and MLDistributions, A/B tests, bias-variance, metrics, leakageExplain each concept from this course in your own words
Case studyFraming a vague business problem with dataPractise: question → metric → data → approach → risks
Project deep-diveYour own work and decisionsKnow every choice in your projects and what you would change
BehaviouralTeamwork, handling mistakes, communicationPrepare short stories: situation, action, result

Questions you should be ready for

  • "How would you measure whether a new checkout page is better?" (A/B test design.)
  • "Your model has 98% accuracy. Is it good?" (Ask about class balance and baselines.)
  • "Explain precision and recall to a sales manager."
  • "What would you do with missing values in income?"
  • "How do you know your model is not leaking information?"

Common mistakes

  • Projects copied from tutorials with no own question or insight.
  • Uploading API keys or personal data to GitHub.
  • Not being able to explain a technique used in your own project.
  • Over-claiming results ("my model is 99% accurate") without context.

Need help?

Stuck on a word or an idea? Search this course, or every course, and jump straight to the explanation or a worked question.

Try:

Watch the lesson

Portfolio, projects and interviews · 1/6

Show, don't list

A portfolio shows employers you can turn messy data into useful answers.

Prefer a video by a teacher? Find free videos on “data science portfolio projects ideas interview preparation” on YouTube ↗

In real life: A project that got interviews

A final-year student analyses public data on school enrolment and teacher numbers by district, finds districts where pupil-teacher ratios worsened, builds a simple model to flag districts likely to cross a threshold next year, and writes a two-page summary with charts and limitations. In interviews, she is asked mostly about this one project, and her clear explanation of data issues and evaluation choices leads to two offers for analyst roles.

Practice (0/2 done)

Task 1 · Practice

Write the README outline for one portfolio project: title, the question, the data source and licence, cleaning steps, method and baseline, results with the main metric, limitations, how to run the code, and next steps.

Task 2 · Code · Python

Interviewers often ask you to compute a metric quickly. Write lift_percent(control_rate, test_rate) that returns the relative improvement of the test over the control as a percentage, rounded to 1 decimal place: (test − control) / control × 100. Return 0.0 if control_rate is 0.

Quick quiz

Question 1

1.Which project is likely to impress an interviewer most?

Question 2

2.What must never be pushed to a public GitHub repository?

Question 3

3.An interviewer says "Your model has 98% accuracy, is that good?" What is the best first response?

Question 4

4.Where can you find Indian public datasets on rainfall and agriculture?

Question 5

5.In a case-study round, what should you do first?

Read more (free)

Open a question to see the answer and why. The first ones are this topic's own quiz, so try the quiz above before you peek.

Which project is likely to impress an interviewer most?

Answer: One project with a real question, clean data, a baseline, honest evaluation and a clear write-up

Depth and clear thinking matter more than quantity.

What must never be pushed to a public GitHub repository?

Answer: API keys and personal data

Secrets and personal data in public repos are a serious risk.

An interviewer says "Your model has 98% accuracy, is that good?" What is the best first response?

Answer: Ask about class balance and the baseline accuracy

Accuracy means little without context.

Where can you find Indian public datasets on rainfall and agriculture?

Answer: data.gov.in, the Open Government Data Platform

It hosts many datasets from government departments.

In a case-study round, what should you do first?

Answer: Clarify the business question and how success will be measured

Framing comes before methods.

In interviews, a strong answer explains a flaw with…Related

Answer: The attack, its detection and its fix

Pairing attack with defence is what interviewers want.

From Building a portfolio and getting hired

Which makes a portfolio project most convincing?Related

Answer: A real problem, tests, a live link and a clear README

Depth and evidence matter.

From Projects, portfolio and jobs

Which portfolio is stronger for a junior full-stack role?Related

Answer: Two or three live projects with clear READMEs and real users

Finished, live projects with good READMEs show that you can make decisions and ship, which clones cannot show.

From Your portfolio and GitHub

Which project usually impresses a recruiter most?Related

Answer: An expense tracker you designed for hostel costs, with your own features, live and with a clear README

A finished, original project that solves a real problem shows real skill and is easy to talk about in interviews.

From Your portfolio project

Pass the quiz (60% or more) or mark this topic as done when you are ready.

Your progress is saved on this device only.

Next: Capstone: an end-to-end project →

Next steps

  1. 1 more topic in “Ethics, privacy and careers” before its exam: Capstone: an end-to-end project.
  2. Module 12 exam: Ethics, privacy and careers · 12 questions · 20 min, without help. 5 topics of this module still to finish first.
  3. 62 topics left in the course before the final exam.

← Previous: Career paths: analyst, scientist, ML engineer, AI engineer