Bloomberg Lp Data Scientist interview typically runs multiple rounds: technical, take-home, and final round. Timeline is months, and the process is notably long with little feedback.
$155K
Avg. Base Comp
$202K
Avg. Total Comp
4
Typical Rounds
2-4 months
Process Length
Our candidates report that Bloomberg is less interested in flashy algorithms than in whether you can work through imperfect, business-shaped data without losing rigor. The technical prompts skew practical: strings, arrays, hash maps, and basic statistics show up alongside sampling, hypothesis testing, error calculations, and descriptive metrics. Just as important, multiple candidates described scenario questions about data quality, trends, and outliers, which tells us the bar is really about whether you can explain what the data is saying when the setup is messy or incomplete.
A recurring theme is that Bloomberg likes exercises with a real-world frame, especially when the answer has to be translated into something a non-technical audience could use. One candidate’s take-home on social vulnerability factors required methodology, calculations, visualizations, and policy insights, and the strongest signal was not volume of analysis but whether the work stayed grounded in the county-level constraint and the census-tract logic. We’ve also seen a round that mixed a simple Python task with a statistics solution and a critique of flaws in the original setup, which suggests they value methodological skepticism as much as correctness.
The non-obvious make-or-break factor here is communication discipline under ambiguity. Even when candidates felt the questions were fair, they consistently mentioned abrupt endings and sparse feedback, so the people who stand out are the ones who make their reasoning easy to follow and their assumptions explicit. In our view, Bloomberg is screening for analysts who can move from code to interpretation to decision support without overclaiming what the data can prove.
Synthetized from 1 candidates reports by our editorial team.
Had an interview recently?
Share your experience. Unlock the full guide.
Real interview reports from people who went through the Bloomberg Lp process.
The hardest part of this process was honestly not the technical questions — it was the amount of time I put in before getting a very abrupt ending. I went through multiple rounds for the Data Scientist role at Bloomberg, and the process started with a 1-hour technical on HackerRank. That round was pretty straightforward in terms of coding difficulty, but it covered a wide spread of basics: data structures like strings, arrays, and hash maps, plus control flow, for loops, and some statistics topics like sampling, hypothesis testing, error calculations, and descriptive statistics. I also got scenario-based questions around data quality, trends, and outliers, so it felt more like checking whether I could reason through messy real-world data than just write code quickly.
The later exercise was much more substantial. I was given a take-home hiring exercise centered on social vulnerability factors in a city, using CDC/ATSDR SVI-style demographic and socioeconomic data. The prompt asked me to limit the analysis to census tracts within one county, then prepare a short presentation or document with methodology, calculations, 2 to 3 visualizations, and 2 to 3 policy insights. It was framed as a roughly two-hour assignment, but it definitely took more thought than that if you wanted to do it well. I also had a round that was split into three parts: an easy Python coding problem, a real-world statistics-based solution, and a taxonomy question where I had to identify flaws in the original setup. Overall, the questions were fair and mostly practical, but the communication was poor throughout. After reaching the final round and spending months on the process, I got an automated rejection email saying they were pursuing other candidates. The interviews themselves were positive, but the candidate experience was frustrating enough that I’d go in expecting a long process and very little feedback.
Prep tip from this candidate
Be ready for a HackerRank-style screen that mixes basic Python with statistics concepts like sampling, hypothesis testing, and error calculations. Also practice turning a messy public-health style dataset into a short, slide-ready analysis with clear methodology, visuals, and policy takeaways, since the take-home was very specific about that format.
Share your own interview experience to unlock all reports, or subscribe for full access.
Sourced from candidate reports and verified by our team.
Topics based on recent interview experiences.
Featured question at Bloomberg Lp
Write a query to return whether each user's subscription date range overlaps with any other completed subscription
| Question | |
|---|---|
| Google Maps Improvement | |
| Longest Increasing Subsequence | |
| Hurdles In Data Projects | |
| Median O(1) | |
| 5th Largest Number | |
| Filling Supermarket Bag | |
| Target Value Search | |
| Binary Tree Validation | |
| Most Repetition | |
| Moving Window | |
| Pathfinder in Maze | |
| Summing Numeric Strings | |
| Shortest Path Algorithms | |
| Check Matching Parentheses | |
| Client Solution Pushback | |
| Minimum Days for Scheduling All Meetings | |
| Why Do You Want to Work With Us | |
| Your Strengths and Weaknesses | |
| Minimum Parking Spots | |
| LRU Cache 1 | |
| Analyzing Multiple Data Sources | |
| Prime to N | |
| Find the Missing Number | |
| Bank Fraud Model | |
| Rectangle Overlap | |
| String Subsequence | |
| Nearest Common Ancestor | |
| Fair Coin | |
| Assumptions of Linear Regression |
Synthesized from candidate reports. Individual experiences may vary.
The process began with a timed technical assessment focused on practical coding and statistics fundamentals. It covered strings, arrays, hash maps, control flow, for loops, sampling, hypothesis testing, error calculations, descriptive statistics, and scenario-based questions about data quality, trends, and outliers.
Candidates complete a take-home analysis centered on social vulnerability factors using CDC/ATSDR SVI-style demographic and socioeconomic data. The assignment asked for analysis limited to census tracts within one county, along with methodology, calculations, 2 to 3 visualizations, and 2 to 3 policy insights in a short presentation or document.
This round was split into three parts: an easy Python coding problem, a real-world statistics-based solution, and a taxonomy question that required identifying flaws in the original setup. The emphasis was on applied reasoning and the ability to critique an analysis design, not just solve algorithmic problems.
The candidate reached a final round before receiving a rejection. Based on the experience shared, this stage appears to have been the last interview step before the automated decision, but no additional details about the format were provided.