Summary
I interviewed for an SDE-2 role at OnlineSales AI, which involved rounds on Data Structures & Algorithms (LRU Cache), System Design (UPI), and High-Level Design (workflow system). I performed well and was ultimately selected for the role.
Full Experience
Round 1 – LRU Cache (DSA)
Interviewer gave a real-world caching example.
The expected solution was an LRU Cache.
I implemented it using Doubly Linked List + HashMap.
We did a dry run, not test cases.
Focus was on logic, structure, and clarity.
Round 2 – Projects + System Design
Discussed my projects, impact, and how I’ve grown as an SDE-2.
Talked about my weekly workflow, ownership, and infra responsibilities.
Then moved into a UPI system design discussion:
End-to-end flow, PIN handling, roles of banks/PSPs/NPCI.
Auth, scaling, services like payment processor and ledger.
Interviewer wanted real understanding, not just theory.
Round 3 – HM + HLD Round
Scenario based :
Amazon is our client and they want to build a workflow like: User adds to cart → wait 30 mins → if no purchase → send notification
Discussed:
How to store workflows (e.g., JSON, state machines).
Use of job schedulers for time-based triggers.
Notification service, retries, and failures.
I did well on event handling, but wasn’t clear on workflow modeling.
Interview ended with a good talk on company culture and look very attached to comapny — very positive.
Verdict - After a few days, I got the call — selected for the SDE-2 role!
Compensation Details: https://leetcode.com/discuss/post/6839557/onlinesales-ai-sde-2-compensation-by-ano-zfok/
Interview Questions (3)
The interviewer presented a real-world caching scenario, and the expected solution was to implement an LRU Cache. I implemented it using a Doubly Linked List + HashMap. We performed a dry run focusing on logic, structure, and clarity.
Design an end-to-end UPI system, including PIN handling, the roles of various entities like banks/PSPs/NPCI, authentication mechanisms, scaling considerations, and core services such as payment processors and ledgers. The interviewer emphasized a real understanding over theoretical knowledge.
Design a system for a client (like Amazon) to implement a specific workflow: When a user adds an item to the cart, wait 30 minutes. If no purchase is made within that time, send a notification to the user.
Discussed how to store workflows (e.g., JSON, state machines), the use of job schedulers for time-based triggers, and aspects of the notification service, including retries and failure handling.