It’s been a whirlwind interview season. The purpose of this post is to document my experience finding a job after my department was laid off in Oct 2024.
Technical assessments
Online
| Platform | Assessments |
|---|---|
| HackerRank | 3 |
| CoderPad | 1 |
| Woven | 1 |
| HireVue | 1 |
| Filtered | 1 |
| Other | 1 |
Topics:
- SQL
- Python
- Base, data structures
- Pandas
- Pyspark
- R
- matrix multiplication
- ML
- Binary classification confusion matrix metrics
- Data analysis
- Behavioral
- Word problems, pattern matching
Take-Home
- Create health status characterizations from provided data
- Python
- SQL, data analysis, and ML
- Python
- Bayesian analysis
- R
- Create a presentation about a past project
Live: in-person or peer-programming
| Focus | Frequency |
|---|---|
| SQL | 10 |
| Python | 5 |
| Data engineering | 2 |
| ML | 2 |
| R | 1 |
| Power BI | 1 |
| Math | 1 |
Data engineering:
- medallion architecture
- start schema
- normalization
SQL:
- simple joins
- complex joins
- primary keys, foreign keys
- indexes
- query optimization
- historical archival to keep only relevant data in table (big data question and how to fix really slow queries)
- data validation
Math:
- identify the function that made the given plot
- integrals and derivatives
- sets
- algebra
One company uses SAS vs Python or R to analyze data…
Presentations
2 panel presentations
- Review of a technical assessment (SQL, EDA, Analysis, ML)
- Overview of a project
Interview questions
Probability
Two friends are playing basketball and practicing free throws. Player 1 has a 75% free throw percentage. While player 2 has a 50% free throw percentage. What is the probability that player 1 scores before player 2.
There are 10 white socks and 10 black socks in your sock drawer. Your room is pitch black. What is the minimum number of socks needed to ensure you have a matching pair of socks?
Given the predicted probability of an event, provided by a binary classification ML model. Calculate at what value at which point the company should or should not act. I was provided financial information and a probability.
Expected return with discrete probabilities of an event. I was provided different scenarios. The final question was a cumulative probability problem in which the goal was to calculate the total cost.
Statistics
- AB testing and experimentation
- how to design an AB test
- how to determine sample size
- what is power, how to calculate it
- how to determine how long to run an experiment for
- MDE - minimum detectable effect
- types of analyses
- hypothesis tests: t-tests, tests of proportions
- diff-in-diff
- propensity scores
- score matching
- inverse probability weighting
- how to analyze the results if the groups aren’t similar (found bias after running the experiment)
- Data analysis
ML
Which is more likely to overfit, random forests or boosted algorithms such as XGBoost?
Solving business problems with unsupervised learning methods (e.g., clustering).
Which loss function would you use for … and why?
One company scheduled an hour long ML technical interview where the interviewer presented a case study and my role was to discuss how I would use ML to solve the problem. The case study pertained to creating an ML data-product to match applicants to jobs. Specifically, which jobs to recommend to applicants via email.
Projects
Lot’s of questions about past projects.
Behavioral
Almost every recruiter asked some form of, “why are you looking for a new job?”
Why do you want to work for company-name? Or, why this role?
Tell me about a time when…
- you used your direct reports to move the needle on a project?
- you had a crucial conversation and provided a direct report constructive feedback
- you mentored a direct report to educate them about something, how did you go about this?
- you conveyed technical information to non-technical audiences
- you communicated the assumptions and limitations of your work to non-technical audiences
Describe an analytical project.
What’s your preferred working style: collaborative, ideating on ideas together or to be instructed on what to do?
How do you explain your frequent job changes? A CTO and a VP have asked me this question. I didn’t have an answer initially, and it cost me a job. So I crafted a response. The second time I was asked this question I was better prepared, but still didn’t nail it. The issue was my answer was all about me, it didn’t take into account the company’s perspective. A good answer I think should explain why I switched jobs and reframe the experiences as a benefit to the interviewer.
Cloud
Azure:
- Which managed services are you familiar with?
AWS:
- Sagemaker
- S3
- Lambda
- Redshift
Databricks
- Are you familiar with
Deep learning
PyTorch and neural nets
Web apps
Questions about building user interfaces using flask, dash, or shiny