Jing demonstrated strong system design fundamentals and clear communication during our session focused on designing a Job Scheduler. While currently aiming for mid-level roles, Jing is actively preparing to meet the expectations of senior-level interviews.

Areas of strength included:

• Proactively aligned on functional and non-functional requirements, checking with the interviewer at each stage
• Enumerated clear and RESTful API designs, including request/response structures
• Presented a structured high-level design, with thoughtful separation between job metadata and execution state (job vs. execution tables)
• Utilized message queues (e.g., SQS) effectively to support decoupling and latency tolerance
• Demonstrated a strong grasp of state machine design, with retry logic and exponential backoff strategies
• Identified valid use cases for DynamoDB, citing its benefits for write-heavy workloads
• Considered complex features like job dependencies, proposing the use of a dependency map
• Deep dive discussion on worker polling strategies and queueing based on delay/execution time showed readiness for real-world challenges

Areas for improvement:

• Expand the discussion on CAP trade-offs, especially in the context of distributed job execution
• Provide quantitative metrics when discussing non-functional requirements (e.g., peak load of 10K jobs)
• Clarify API responses — avoid returning large object payloads; instead, return a simple job_id
• For better realism, discuss API contract design in cases like job creation (e.g., how job_id is returned post-persistence)
• Reinforce the rationale for DynamoDB by connecting to scaling under burst loads and high throughput updates
• Elaborate on the complexity of dependency graphs, possibly suggesting a separate table or use of a graph database
• Mention indexing/caching layers explicitly to support low latency retrieval
• Explore implications of "at-least-once" delivery semantics, and how retries can be made idempotent

Suggestions for continued practice:

As a next step, I recommend preparing for questions like “Design a Distributed Key-Value Store” — a natural progression from the job scheduler, with deeper focus on partitioning, consistency, and replication trade-offs.

Jing has shown promising potential. With more mock practice and deliberate work on advanced system design discussions, I’m confident they’ll be much more prepared for interviews at companies like Snowflake and TikTok.