A Data Scientist Interview Experience

Two months ago, I received a rejection after Stage 2 (the Data Cleaning exercise) at a top-tier hedge fund. I wanted to share my overall experience reflecting on this journey. Over the past few months, I have been applying for various roles across different companies in Hong Kong. When I initially received the email from this firm, I was pleasantly surprised. Admittedly, I didn't know much about them at first. After some quick research, I realized they are a top-tier player in the industry.

Initially, the recruiter reached out for a quick 10-minute introductory call. Although we couldn't connect due to a scheduling conflict on their end, the recruiter proactively emailed me a comprehensive breakdown of the 4-stage interview process. I really appreciated this clarity, as it gave me a transparent roadmap of what to expect.

DATA SCIENCE & ALGORITHMS

Stage 1 consisted of one SQL problem, one Data Science (ML training) problem, and one Python problem; these problems needed to be finished within one hour. During my preparation days before the actual assessment, I solved the preparation toolkit problems on HackerRank. I observed that I was able to answer easy coding problems, but I found medium and hard ones really challenging. During the actual Stage 1 assessment day, I was able to pass it. It was a great learning experience. The lesson I learned is the value of solving problems on the HackerRank platform; it is a useful website to learn and grow in algorithms and coding. As a non-CS major during my undergraduate years, HackerRank is a great platform to grow daily in algorithms and coding concepts. As my next steps, I will allot 20–30 minutes every day to solving problems on HackerRank. I learned from this Stage 1 process that practice makes perfect, and that I can do it as long as I persist.

DATA ANALYSIS

Stage 2 was a data cleaning exercise. They gave me a CSV file and instructions to clean the data, which included columns like Date, Open, High, Low, Medium (Price / Tier), and Close. The objective was to properly clean the data and explain whether the signals provided were useful. While doing the task, I spent many hours finishing it and made sure it was done rigorously. I worked on it in the Hong Kong Central Library. For the data analysis, I used AI to help analyze the dataset while making sure that I understood the critical steps. As a beginner, there was probably something wrong with my quantitative analysis. I suspect there was an issue with the way I processed the signals or cleaned the data that did not align with proper alpha research. What I learned from this process is to be more rigorous with my steps and approaches, making sure every step is approached with rigor and careful scrutiny.

WHERE I FAILED

PRESENTATION: I think I failed in the presentation of the results. I documented the results in Notion and took some screenshots of the code terminal (via pdb). This was not very ideal. It would be more ideal to use VS Code with Markdown and code for a more professional presentation of results

LACK OF DOMAIN KNOWLEDGE. Before this interview, I did not have any domain expertise to understand trading. Personally, I do not trade stocks. Everything was new to me; I did not really understand how to process a time-series dataset of stock signals. I did not fully understand how trading works.

CONCLUSION

While this data scientist interview didn't end with an offer, I view this setback as a valuable opportunity to refine my data science and analysis skills. I learned an immense amount throughout the process and am incredibly grateful for the experience. A huge thank you to the recruiting team for being so supportive along the way! To prepare for future interviews with alpha trading firms, I will apply the lessons learned from this experience. Moving forward, I plan to deepen my knowledge of alpha generation strategies and improve my understanding of the entire alpha mining process.