"Tell me about yourself" is the first question in almost every interview.
Many people treat it as small talk. That's a mistake. It's not a test — it's an icebreaker and an anchor. The interviewer uses these two minutes to decide where to dig next. So don't answer passively: script the rest of the interview yourself.
Three Hard Rules
- Two minutes max — about 220–270 words in English. Going over costs you precious evaluation time.
- Lead with impact, not responsibilities — "cut inference latency by 40%" beats "responsible for optimizing models."
- Make every section "clickable" — frame each experience like a link: give enough context to spark a follow-up, but don't dump everything. Save the depth for the Q&A.
The Framework: Present → Past → Future
| Section | What to say | Time |
|---|---|---|
| Present | Who you are, your current role, one standout metric | ~30s |
| Past | One or two most relevant projects, told through results | ~60s |
| Future | Why this role / this company | ~30s |
💡 Key insight: For tech interviews, lead with the present. Let the interviewer place your current level first, then go backwards — don't start from college.
Plug-and-Play Template
Replace the [brackets] with your own content.
"I'm a
[title, e.g. machine learning engineer]with[X]years of experience, currently at[company/team], where I[one line on what you do].Most recently, I
[most relevant project]—[what you did], which[quantified result, e.g. cut inference latency by 40%]. Before that, I[second experience in one line].What draws me to
[company]is[specific reason tied to the JD]. I'm especially excited about[a product/tech direction of theirs], and my background in[your strength]should let me contribute quickly."
Worked Example (fictional data, for tone and pacing only)
"I'm a machine learning engineer with about three years of experience, currently at a fintech startup where I build and deploy fraud-detection models that score around two million transactions a day.
The project I'm most proud of: our fraud model had good offline AUC but kept misfiring in production. I rebuilt the feature pipeline to kill training-serving skew, added real-time features, and ran an A/B test — that cut false positives by 35% while keeping recall flat, saving the ops team a huge amount of manual review. Before that, I was more on the data side, building the labeling pipeline the team still uses.
What draws me to your team is that you're solving fraud at a much larger scale and you publish real research on it. I'm especially excited about your work on graph-based detection — my background in feature engineering and production ML should let me contribute quickly."
Polish Checklist
- Say it out loud and time it — cut the second experience if you go over two minutes
- Map every "clickable" hook to a STAR story you're ready to expand on
- Don't open with "I'm trying to transition…" or "I don't have much experience…" — lead with what you can do now
- Always tie the Future section to the company's JD / product — no generic filler
- Use the same structure in both languages so it's easy to memorize and stays consistent
Wrap-Up
A self-introduction isn't an autobiography — it's a carefully designed opening line. Remember three things: lead with the present, use numbers, leave hooks. Practice these two minutes until they roll off naturally, and you own the rhythm of the whole interview.