On 8 September 2026 Qube Research & Technologies posted its 2027 QR & Trading internship. The form asks for an 80 to 100 word written answer and tells you to apply to one location only. After that, the posting says, comes "a coding challenge focused on core programming and problem-solving skills".
That is roughly all QRT says in public about how it interviews. The rest has to be pieced together from candidates.
Here is the short version. The QRT interview is an application, then a coding assessment, then somewhere between two and five interviews held on site or over Microsoft Teams. Research candidates are tested on statistics, machine learning, probability and a deep dive into their own projects. Developer candidates get LeetCode-style problems at easy to medium level, plus questions on Python or C++ and computer-science fundamentals. The number of rounds varies by role, office and team, and no two candidate reports describe quite the same loop.
For the firm itself (history, strategies, culture), see our Qube Research & Technologies guide. This page is about getting through the door.
The Process by Track
QRT's own postings describe the stages for each programme. What happens inside those stages comes from candidates.
| Track | Official stages | What candidates add |
|---|---|---|
| QR & Trading internship | Application with a short written answer, coding challenge, interviews on site or on Teams | Two to four technical rounds, heavy on statistics and ML, with a project deep dive |
| Software Engineering & Quantitative Development internship | Application, coding challenge, "Assessment Centre / Online Interviews" | LeetCode easy to medium, Python or C++, data structures |
| QR PhD & Postdoc | Application, then interviews on "technical expertise and alignment with our collaborative culture" | Deeper research discussion; candidates are invited to try a data challenge |
| FPGA (Hong Kong) | The assessment is a technical interview with the team | Few reports |
QRT does not name its assessment platform or say how many interview rounds to expect.
The Application
Two details in the 2027 postings are worth taking seriously.
The written answer is short, 80 to 100 words, and QRT says it "read[s] applications carefully, so thoughtful answers… are useful". A generic paragraph about loving markets wastes the only space you get to sound like a person. Say what you have built or studied and why it points at QRT's kind of problem.
You also apply to one location. Reviews are rolling, so there is no advantage in waiting for a deadline, and QRT says it will "consider your skills and interests alongside the problems our teams are working on". In other words, team matching happens during the process, not after the offer.
The Coding Assessment
Officially, it is a coding challenge on "core programming and problem-solving skills". Candidates fill in some detail:
- Hong Kong quant researcher, December 2025: a HackerRank test with three LeetCode-medium problems.
- London quant developer intern, February 2025: two problems at LeetCode easy to medium.
- Reddit, 2024 and 2025: some research candidates got an online assessment first, others went straight to a technical interview.
Some prep sites say QRT uses Codility. We found no candidate who said so directly, while one named HackerRank. Treat the platform as unconfirmed and practise timed algorithm problems in whichever of Python, C++ or C# you are strongest in, since QRT lists all three. Our Python quant interview questions are a good place to start.
Research Interviews
The research loop is where QRT differs most from prop trading firms. None of the reports we found mentions a market-making game. Candidates describe statistics and machine learning first, probability second, and a long look at their CV.
Recent reports, by office:
- Hong Kong, December 2025: after the coding test, a technical interview on Python and market intuition. The whole process took under a month.
- London intern, February 2026: three technical rounds over Zoom with a deep dive on projects. Topics included coin-toss statistics, z-scores and p-values, regression and penalisation, and ML questions on correlation and high dimensions.
- London, April 2025: a first interview with a portfolio manager, then past experience plus "hard questions on random forests".
- Paris intern, August 2025: a phone interview mostly on statistics, including t-statistics and basic ML algorithms.
- Zurich intern, November 2024: a theory interview on ML, then a two-week take-home ML project, a remote interview on the solution and a longer onsite. This is the only report of a take-home we found.
Random forests and linear regression each come up in more than one report. One Glassdoor reviewer says "Markov chains questions are a favourite".
Worked Questions
The first two are reported by candidates. The rest are our own examples in the style candidates describe, and we have labelled them that way.
1. Two regression slopes (reported, Hong Kong). The slope of y on x is b1 and the slope of x on y is b2. What is the relationship?
b1 = Cov(x, y) / Var(x) and b2 = Cov(x, y) / Var(y). Multiply them and you get Cov² / (Var(x)Var(y)), which is the squared correlation. So b1 × b2 = ρ². The two slopes always share a sign, their product is at most 1, and b2 equals 1/b1 only when the fit is perfect. The common wrong answer is that one slope is the reciprocal of the other.
2. Counting paths to a position (reported, Hong Kong). You start flat and want to end long 10 of A and short 10 of B. Each step, you either buy one A or sell one B. How many paths are there?
Every path is 20 steps, 10 of each kind, so you are choosing which 10 of the 20 steps are A trades: C(20, 10) = 184,756. The candidate also mentioned a harder variant with four possible moves per step, but did not give its exact rules, so we will not guess at them.
3. Zero correlation without independence (reported in an older interview). Give an example of two variables with correlation zero.
Take X uniform on [-1, 1] and Y = X². Then Cov(X, Y) = E[X³] - E[X]E[X²] = 0 - 0 = 0. Y is a function of X, so they are as dependent as variables can be, yet the correlation is zero. Correlation only measures linear association.
4. Is the coin fair? (our example, in the reported style). You flip a coin 100 times and get 60 heads. Is it fair?
Under a fair coin the count has mean 50 and standard deviation √(100 × 0.5 × 0.5) = 5. So 60 is two standard deviations out, z = 2, and the two-sided p-value is about 0.046. With a continuity correction, z = 1.9 and p is about 0.057, which is close to the exact binomial answer. The honest answer is "borderline": it sits either side of 5% depending on the method, which is exactly the point to make out loud.
5. Why random forests work (our example, in the reported style). Why does averaging many trees reduce error, and why subsample features?
If each tree's prediction has variance σ² and any two trees have correlation ρ, the average of B trees has variance ρσ² + (1 - ρ)σ²/B, a result set out in Hastie, Tibshirani and Friedman's The Elements of Statistical Learning (2009). Adding trees kills the second term but never the first. Feature subsampling lowers ρ, which is the only way to shrink the part that more trees cannot touch.
6. Ridge or lasso (our example, in the reported style). When would you choose one over the other?
Ridge (an L2 penalty) shrinks all coefficients towards zero and handles groups of correlated predictors by sharing weight between them. Lasso (L1) sets some coefficients exactly to zero, which gives you feature selection but can pick one predictor from a correlated group more or less arbitrarily. With many weak, correlated signals, which is typical of financial data, ridge or an elastic net is usually the safer start.
For more practice across these topics, see our quant research interview questions and probability interview questions.
Developer and Engineering Interviews
Developer candidates describe a more conventional software loop, with a finance flavour.
- London quant developer intern, February 2025: two coding problems in the assessment, then a behavioural and technical round with a LeetCode medium, then a final round with a medium-to-hard problem plus Python and data-structure questions. It took one to two months.
- Quant developer, July 2025 (Glassdoor): about two months, with interviews across several teams. Questions centred on the CV and computer-science fundamentals: data structures, syntax, multithreading and databases, at LeetCode easy-to-medium level.
- C++ developer, April 2025 (Glassdoor): algorithm questions, C++ template programming and behavioural questions.
- Low-latency role, Amsterdam (Reddit, March 2026): a two-hour onsite.
QRT's 2027 developer internship asks for C++, C# or Python, plus Git, SQL, GCC and Linux, and says it wants people "interested in building things, not just studying them". Personal projects and open source count, in QRT's words, "wherever you gained it".
Our quant developer interview questions and C++ quant interview questions cover this loop. If you are still deciding whether the route suits you, start with how to become a quant developer.
The Data Challenge Route
This is QRT's most unusual door, and it is official. Research postings invite candidates to enter QRT's data challenge on the ENS Challenge Data platform, and say "strong performance may lead to direct follow-up". The 2026 challenge, on asset allocation performance forecasting, opened on 22 January 2026.
A good leaderboard finish is concrete evidence of exactly what the research interviews test: careful validation on noisy data. It is not a substitute for the application, and QRT does not say what rank earns a follow-up.
2027 Programmes
At the time of writing (29 September 2026), QRT's jobs board lists these early-careers programmes:
- QR & Trading internship, Europe and Middle East. Aarhus, Budapest, Dubai, Geneva, London, Paris or Zurich. Four to six months in 2027, for penultimate or final-year bachelor's, master's or PhD students.
- QR & Trading internship or graduate role, Asia. Hong Kong, Singapore, Shanghai or Beijing, for three to six months.
- Software Engineering & Quantitative Development internship. Software engineering in London, Wrocław and Zurich; quantitative development in London, Paris, Wrocław, Zurich and Dubai. Three to six months, or a 12-month industrial placement. Candidates should be on track for a 2:1 or above.
- QR PhD & Postdoc. London, Paris and Zurich, for final-year PhD students and postdocs.
- Others. Data engineering, FPGA, infrastructure and security internships in selected offices.
Postings change, so check QRT's careers page before you apply. General questions go to campus@qube-rt.com.
Timelines and Silence
Glassdoor users put the average time to hire at about 22 days across 34 submissions (as of July 2025), with quant research internships the slowest at about 39 days. Individual reports on Wall Street Oasis range from under a month to three months.
The less pleasant pattern is silence. In one Reddit thread from early 2026, four candidates described waits of one to three weeks or more after late rounds, and no formal rejection. If you are holding another offer, tell your recruiter the deadline early.
Where This Guide Goes Wrong
Almost everything about QRT's interview content comes from a few dozen anonymous reports across Glassdoor, Wall Street Oasis and Reddit. The loop clearly differs by office and team: the Zurich take-home and the Hong Kong market-intuition round may say more about one team than about QRT.
The firm is also growing and changing fast. It is reported to be building discretionary equity and macro teams, whose hiring will not look like the systematic research loop described here. And the questions we wrote ourselves are illustrations of the reported topics, not leaked questions.
Recruiting Notes
Nothing here guarantees an assessment format, an interview or an offer. Formats and round counts are candidate-reported and change between cycles. For pay, our QRT salary guide collects the public data points; every figure in it is an estimate, not a number provided by QRT, and packages vary by role, office, year and performance.
Frequently Asked Questions
How many interview rounds does QRT have?
It varies. QRT's postings describe an application, a coding challenge and then interviews, without a fixed count. Candidates report anywhere from two to five rounds after the coding test, depending on the role, office and team. One Reddit account describes one technical interview with a researcher followed by three onsite interviews.
What is the QRT online assessment?
A timed coding test. QRT describes it as "a coding challenge focused on core programming and problem-solving skills". Candidates report two or three problems at LeetCode easy to medium level, and one names HackerRank as the platform. QRT does not confirm the platform.
Is the QRT interview hard?
Glassdoor users rate it about 2.9 out of 5 for difficulty. The research rounds are demanding in a specific way: they test statistics, ML and how well you can defend your own projects, rather than speed arithmetic or trading games.
What questions does QRT ask quant researchers?
Mostly statistics and machine learning: regression and penalisation, random forests, z-scores and p-values, correlation. Probability questions such as Markov chains and dice problems also appear, alongside a detailed discussion of your CV projects.
Does QRT hire undergraduates for quant research?
Yes. The 2027 QR & Trading internship is open to penultimate and final-year bachelor's students as well as master's and PhD students. The QR PhD & Postdoc track is separate.
What is the QRT data challenge?
A public machine-learning competition QRT runs on the ENS Challenge Data platform. QRT's research postings encourage candidates to enter and say strong performance may lead to direct follow-up. The 2026 challenge, on asset allocation performance forecasting, opened on 22 January 2026.
Practise the questions QRT Interview: Process, Questions and 2027 Internships actually asks
Reading about the interview is one thing - sitting one is another. Open your free Quantt prep workspace for a real course lesson plus interview-style coding tests modelled on firms like Jane Street, Citadel, Hudson River and Optiver.
Free lesson + interview practice · No credit card required