Technology Track

Data & Machine Learning Interview Prep

The modelling questions are the easy half. What separates candidates is whether they can say how a model fails, who it fails for, and how they would know before a customer does.

By Surojit Chakraverti, founder of L3VLUP and an investor running a long-short healthcare and technology equities strategy.

Start Preparing View All Career Paths

1,000+ candidates coached into top-tier firms

What We Cover

SQL beyond joins: window functions, cohorting, and the queries interviewers actually set

Statistics that come up in the room: hypothesis testing, confidence, power, and multiple comparisons

Framing a problem, which is where most interviews are won and lost before any modelling starts

Model selection and the trade you are making, rather than reaching for the most complex option

Leakage, drift and the failure modes that only show up after deployment

Experiment design and A/B testing, including the tests that should not be run

Communicating a result to somebody who will act on it and cannot read your code

What is different in finance: non-stationarity, regime change, and why a backtest flatters you

Common Data & ML Interview Questions

Walk me through how you would detect fraudulent transactions on a payments platform.

Your model performs well offline and badly in production. Where do you look first?

Write a query returning each customer’s first and third purchase date.

How would you decide whether a change to the ranking algorithm actually helped?

What is data leakage, and give an example you have actually caused or found.

When would you prefer a logistic regression to a gradient boosted tree?

How do you evaluate a model where the positive class is 0.1% of the data?

A stakeholder wants a model for something you think should not be modelled. What do you do?

These are real questions asked in data & ml interviews. Our coaching covers how to structure and deliver winning answers.

What Top Firms Look For

Framing before fitting: candidates who ask what decision the model supports before choosing an algorithm

Honesty about uncertainty, including saying that a result is not significant when it is not

Awareness that a model in production is a system with failure modes, not an artefact with an accuracy score

SQL that is genuinely fluent, because it is the tool most used and most often overstated on a CV

The ability to explain a method to somebody who will make a decision on it

A Day in the Life: Data & ML

Less modelling than the job title implies. Most days involve understanding what a question is really asking, finding out whether the data can answer it, and discovering that it cannot in the form requested. Model building is a minority of the week; the majority is data work, validation, and conversations with the people who will act on the output. In finance the added constraint is that the world moves: a relationship that held for two years can stop holding, and noticing that quickly matters more than the original model was clever.

How to Break Into Data & ML

Get properly fluent in SQL. It is the most tested and most overstated skill on a data CV

Build one project end to end, including the unglamorous parts: data collection, validation, and what you did when it broke

Be able to explain a project to a non-technical listener in two minutes without using the word model

Learn the failure modes — leakage, drift, imbalanced classes — because interviews probe them and courses skip them

For finance seats, understand why time series break the assumptions most tutorials rely on

What it pays

On the data science & ml ladder, a entry sits at a midpoint around £60k in cash and a principal / head of near £190k. Equity is missing, and in London it matters at the scale-ups and the US firms’ London offices rather than everywhere. The wider point the national data makes: the UK-wide median for data analysts is £38,572 and for programmers £56,914, so the London bands below describe a market well above the national one, not the country.

Data science & ML pay, level by level

Guides for this track

The written material behind the coaching, free to read.

Data Science Interview Questions in Finance9 min read

Frequently Asked Questions

Do I need a PhD for data science or ML roles?

For applied roles, usually not. For research positions at labs and some quantitative funds, often yes. The larger determinant for most jobs is demonstrable end-to-end work rather than credentials, and interviews are increasingly structured to find out whether someone has shipped something or only trained something.

How much SQL do I actually need?

More than most candidates prepare for. Window functions, cohort queries and self-joins appear regularly, and SQL is tested precisely because it is hard to bluff and easy to overstate on a CV. It is the single highest-return thing to drill before an interview.

Is data science in finance different from data science in tech?

The methods overlap heavily; the constraints do not. Financial data is non-stationary, the signal-to-noise ratio is far worse than in most consumer problems, and a relationship that held historically can stop holding without warning. Interviews at funds and banks probe that awareness directly, and candidates from a pure tech background are often caught by it.

What is the difference between a data scientist and an ML engineer?

Roughly, whether the output is a decision or a system. Data science leans towards analysis and inference for someone else to act on; ML engineering leans towards models running in production, with the software engineering that implies. Titles are used loosely, so read the responsibilities rather than the header.

This week in your target role

Open NewsViz

“Tell me about something happening in this market” is coming in some form. Pick one story, read the source, and have a view on it.

Your Coach: Surojit Chakraverti

Citigroup · Rothschild & Co · Hedge Fund CIO

Surojit is a hedge fund CIO (Third Wave Capital, previously Redline) and former investment banker who worked at Citigroup and Rothschild — with his core deal experience in M&A at Citi and Rothschild. He has personally coached over 1,000 candidates into roles at Goldman Sachs, Blackstone, KKR, Google, McKinsey, and dozens of other top-tier firms.

Every data & ml coaching session is tailored to your specific targets, timeline, and experience level. No generic advice, just the insider knowledge that separates offers from rejections.

Ready to Ace Your Data & ML Interview?

Get 1:1 coaching from professionals who have been through the process and sat on the other side of the table. No generic advice, just what actually works.

Book Your Session

Coached by Surojit Chakraverti — Citi, Rothschild, Third Wave Capital

Interview Prep →

Consulting

Interview Prep →

Tell me when Data & ML roles open

Most of these assess on a rolling basis, so the fortnight after applications open is worth more than the deadline printed at the end of it. We watch the boards daily and email you the day one goes live.

One email when applications open, plus a short weekly round-up in season. Free, and you can leave at any point.