Stop 15 · Interview PreparationBehavioral · 1/4

Explain a Past ML Project

Part of the Data Science Roadmap

In short

'Walk me through a project you're proud of' comes up in almost every interview. You know it's coming, so there's no excuse for a rambling answer. Prepare a tight two-minute story.

Walking an interviewer through a real project (from your portfolio or work) — the problem, your approach, what you found, and its impact — one of the most predictable interview questions.

The mascot sits on a park bench with a cricket bat, acting out a shot for a friend.

Think of it like this

Retelling the match

When a friend missed the match, you don't list every ball. You say: we needed 40 off 3 overs, I came in at number six, hit two sixes off the spinner, and we won with a ball to spare. Situation, your role, what you did, the result. A project story needs that shape.

Watch it happen

The two-minute project story

Problem
Your role
Approach
Result
Learning

1/5One or two lines: who had the problem and why it mattered. 'Our hostel mess wasted a lot of food daily.'

The big ideas

What you need to know

Two minutes, then zoom in

Give the short version first. Let the interviewer pick what to dig into.

Own your part

Interviewers are hiring you, not your team. Be specific about your decisions.

Numbers with a baseline

'Error dropped from 18% to 11% vs the old rule' beats 'it worked well'.

Show one trade-off

Why you chose a simpler model, or fewer features, tells them how you think.

How it works

Preparing your story

  1. 1

    Pick one project

    Choose one with a real problem, clear result and decisions you made yourself.

  2. 2

    Write the script

    Five short parts, about 250 words in total. Read it aloud and trim.

  3. 3

    List follow-ups

    Write the five questions you'd ask if you were the interviewer.

  4. 4

    Rehearse on video

    Record, watch once, fix filler words and the slowest part.

Save yourself the pain

Common mistakes

Oops

Opening with a list of tools: 'I used Python, pandas, XGBoost, Docker...'

Do this instead

Open with the problem and why it mattered. Mention tools only when they explain a choice.

Oops

Ending with 'and it worked pretty well'.

Do this instead

End with a number against a baseline, the impact, and one lesson.

Your mission

Script and record

~30 min

Quick check

The interviewer asks, 'What would you do differently?' Which answer is strongest?

Go deeper

  1. 1Prepare a tight 2-3 minute walkthrough of your strongest Phase 14 portfolio project.
  2. 2Structure it as: problem, approach, modeling choices, results, impact — in that order.
  3. 3Practice fielding a technical follow-up question about your methodology or trade-offs.

More behavioral at this stop

Stuck on this topic? Ask an Insider

Get 1:1 guidance from people who've walked this exact path — on the InsideEdge app.

Download
InsideEdge