Interview Process

1. Third-party recruiter

  • They only want to acquise as much as candidates as possible.

  • They are only payed when you accepted the offer.

  • So if you have another offer, or request higher salary they do not accept because the higher risk that you do not accept their offers.

=> Do not talk anything about salary expectations, job searches with them.

2. Internal recruiter

  • They schedule the slots for you to interview round.

=> waste time and cost of company => must be impressed them.

=> Say as much keyword in JD as possible so that they process you.

3. Live Coding

  • 2 mins -> clarify requirements, input/output/contrains.

  • 3 mins -> Come up with draft solutions, why do you do in steps: why sorting, why looping, why variables.

  • 5 - 10 mins -> Implement the solution and explain each steps.

=> First impressive when you finish solution in 10 - 15 mins => Show you are good problem-solving skills.

=> Other 30 mins, they use to struggle you => Focus on thinking process, and completed it is important => Practice optimization skills.

=> At least 2 - 3 solutions.

4. System Design Interview

4.1. High-level design:

  • Functional requirements: Core features (send message, post tweet, search user, etc.).

  • Non-functional requirements: Scale, latency, throughput, availability, consistency.

  • Traffic estimates and capacity planning: Requests/sec, storage size, growth projections.

  • Architecture components:

    • Load balancers
    • Backend servers
    • Databases (SQL vs NoSQL, sharding, replication)
    • Caches, queues, CDNs, pub/sub

  • Data flows: Read path vs write path, how requests move through the system

  • Trade-offs: CAP theorem, eventual vs strong consistency, cost vs scalability.

4.2. Low-level design:

  • Object-oriented modeling: Classes, objects, interfaces, relationships

  • Design patterns: Factory, Strategy, Observer, Singleton, etc.

  • APIs and methods: Clear function signatures, parameters, return types

  • Data models: Table schemas, indexes, relationships, how queries are optimized

  • Component interactions: Sequence diagrams, how modules talk to each other

  • Edge cases and validations: Error handling, retries, testing considerations

  • Maintainability: Clean abstractions, modularity, extensibility

4.3. Notes

  • Error 1: Do not ask the requirements, e.g. number of active users, scalability.

  • Error 2: Do not talk in order of logic, e.g. talk about high-level design first, talk database design later => You need systematic logic to strange system.

  • Error 3: Estimate 1000 images, each image 1 MB => Need to store 1TB => So we need sharding.

  • Error 4: Detail each fields in table, what fields should be index.

  • Error 5: Each level different, require the structure to answer different -> Why to choose the technology rather than other technology.

  • Error 6: Use the techstack that the interviewer know.

  • Error 7: Show knowledge about Redis, Kafka, Event-Driven Design, Serverless.

  • Tips to learn:

    • High-level design research about architecture patterns: Microservices, Distributed System, Event-driven, Serverless.

    • Read each components.

    • Database Design, API Design, Sequence Flow.

=> Senior and Mid is same process, different knowledge dive deep, Senior ask for How Redis fast, How Kafka fast rather than Mid Level.

  • Error 8: Choose the architecture that you know best => Show the trade-offs about that you know about this architecture.

  • Error 9: Migrate system => You change its components, rollout later.

  • Error 10: Operational Mindset -> Deploy, rollback, versioning, monitor.

  • Error 11: Technical adjustment when the requirements changed.

  • Error 12: Technical fundamentals is not changed but trade-offs are unlimited.

  • Error 13: Link the design with pass experiences => show that you have real experience but not copying from another posts.

4.4. Mock Interviews

  1. Functional requirements: based on skateholders.

  2. Non-functional requirements: RPS, QPS, number of users, latency (400ms P99), availability (master-slave)

  3. API Design: POST (driver), GET (user)

  • Endpoint

  • Authenticate

  • Request: timestamp to later request

  • Response

  • API version

  1. Database Design:

  • Do not need to store history

  • Used field last_update_time to do not overwrite update.

  • Do not have relationship, used NoSQL Database: DynamoDB, MongoDB,..

  1. High-level design:

  • Client App: what it have.

  • API Design: What it do.

  • Location Service: What it do.

  • Location DB: What is store.

=> Handle write-heavy: Using queue, route by driver_id.

=> Notes: Check non-functional requirements.

=> Notes: Use cache to update the latest location from cache but not query db.

  1. Failure case

  • Retry.

  • Back propgation.

  • Skip it all.

  1. Total requests:
  • Shard by region.
  1. Network:

  • Using Websocket.

  • Compare HTTP Pooling.

4.4. Senior+

  • Focus soft skills: leadership, ownership, skateholders, communication and convinction cross-team.

5. Lesson learned

  • Sky Mavis Interview: Be careful in implement, dry run the source code, do not easily to conclude when the problem is similar.

  • Senior Grab Interview: Clarify problem more detailed, explain reason for clearly.

  • Middle Grab Interview: Optimize the code to O(N).

6. Resumes

  • Used to impress the recruiter.

  • Write it shortly in each bullet and focus more in impacts.

7. Hiring Manager

  • Decide whether to fill the headcount.

8. How to search jobs

  • When you email a real human and they don’t respond, that hurts: you put yourself out there, someone made a value judgment about you, and you lost.

=> You need to sale yourself.

  • Build personal brands or open-sources.

  • Email EM or Linkedin Sale Navigator for opportunities.

Last Updated On November 10, 2025