hudson | Hudson River Trading | Interview Experience
Interview Date: Not specified
Result: Not specified
Difficulty: Not specified
Interview Process
I applied for the Algorithm Development Intern position for Summer 2026 based in Singapore. After completing the online assessment (OA), I had an interview that consisted of two rounds. The first round focused on probability questions, where the coderpad was used primarily for running simulations related to probability theory. The second round involved live coding.
The interview began with a review of my resume, where we discussed some of the projects listed. We then transitioned into the probability topic. The problem presented was set in the context of a trading company, where we receive profit (X_t) for each time index (t). We considered (n) days, from (X_1) to (X_n), and were interested in the average return. The interviewer asked which metric—mean or median—I would choose to compare.
A specific scenario was given: with a probability of 0.2, we set (X = \mu); with a probability of 0.8, we set (X \sim \mathcal{N}(\mu, 1)). I was asked to determine whether it was better to estimate (\mu) using the mean or the median in this case, and I could use numpy in the coderpad to simulate.
Next, I was asked to mathematically prove that (P(|\hat\mu_{\mathrm{median}} - \mu| > 0.1) < P(|\hat\mu_{\mathrm{mean}} - \mu| > 0.1)), demonstrating that using the median yields a smaller probability of exceeding a 0.1 error margin. We then discussed intuitively why this situation occurs, including when it might be better to use the mean versus the median.
There was a question-and-answer segment at the end. Overall, I found the interview experience to be very positive! It was challenging, but the interviewer provided appropriate hints and support throughout.
Technical Questions
- Probability - Estimation of (\mu) using mean vs. median
- Mathematical proof of error probabilities
Tips & Insights
- Be prepared to discuss your projects in detail.
- Understand the implications of using different statistical metrics in various scenarios.
- Practice simulating scenarios using coding platforms like numpy.