Oracle | SMTS | Interview Experience
Summary
I interviewed for a Backend + Data SMTS role at Oracle Health, completed five rounds—including screening, system design, coding, and behavioral—and received a positive outcome.
Full Experience
YOE: 4 years Role: Backend + Data SMTS at Oracle Health
Round 1 - Screening
There were a huge number of URL access logs, and I was asked to determine the top K URLs overall and the top K URLs accessed in the last 24 hours.
Given that the data was huge and RAM was limited, the interviewer was mainly interested in the approach and trade-offs. I was asked to write pseudocode for the solution.
Round 2
Given a stream of events in the format (userId, eventType, timestamp), I had to determine the top 100 users based on event count in the last 1 hour.
The interviewer first asked me to design an HLD and probed on batch vs streaming processing, Flink vs Spark Streaming, handling the sliding time window, scalability and data volume. After the HLD discussion, I was asked to code the main methods for the solution, similar to LLD.
Round 3 I was asked a DSA problem.
Find the maximum sum of any contiguous subarray having length ≤ K.
After that, the interviewer went through my projects and asked questions based on my resume.
Round 4 - Hiring Manager
Design and implement a system that ingests microservice logs and returns the top K microservices with the highest number of error logs in the last 5 minutes.
The discussion started with HLD, where we covered NFRs such as scalability, availability, fault tolerance and handling high-volume logs. I was then asked to code the main methods for solving the problem. The round ended with a few behavioral questions.
Round 5 - Behavioral This round was primarily focused on behavioral and situational questions based on my background. Since the interviewer also had a technical background, some of the situational questions involved deep dives into the technical details of my past work and projects.
Overall, the interview experience went well. After the pre-screening round, the remaining interviews were conducted in a loop format over 2 days — 2 interviews per day. One thing I liked about the loop interview format is that the final decision is based on a holistic evaluation across all rounds, rather than interviews being eliminatory.
Interview Questions (4)
Top K URLs overall and in last 24 hours
Given a huge number of URL access logs and limited RAM, determine the top K URLs overall and the top K URLs accessed in the last 24 hours. The interview focused on the approach, trade‑offs, and pseudocode implementation.
Top 100 active users in the last hour
Given a stream of events in the format (userId, eventType, timestamp), compute the top 100 users based on event count in the last 1 hour. The discussion included high‑level design choices such as batch vs streaming, Flink vs Spark Streaming, and sliding time window handling, followed by coding the main methods.
Maximum sum subarray with length constraint
Find the maximum sum of any contiguous subarray whose length is less than or equal to K.
Design log ingestion system for top K error microservices
Design and implement a system that ingests microservice logs and returns the top K microservices with the highest number of error logs in the last 5 minutes. The interview covered high‑level design, scalability, availability, fault tolerance, and then required coding the core methods.