How Would You Scale Your Take-Home to 10 Million Users?

Question

Looking at your take-home submission, walk us through how you would evolve it to handle 10 million active users with 50K concurrent API requests.

What this round tests

• Can you critique your own work?
• Do you understand the gap between a working prototype and a production system?
• Architectural thinking: statelessness, caching, async processing

Common areas probed

1. How would file parsing scale? (async job queue, S3 + worker)
2. Where would your DB become the bottleneck?
3. What would you cache, and what's your invalidation strategy?
4. How would you handle partial failures in the pipeline?
5. What observability would you add?

Tip

Be honest about shortcuts you took. CRED engineers appreciate intellectual honesty over defensive justification.