Apollo - EM role

I had my first round as system design, asked to build the live commenting system on live streams.
Functional Requirements

1. Able to live comment
2. User can see the past hisotry as well when joins the live stream
3. 7 day history
4. Abuse words starmarked(extension)
5. Top Comments

Non Functional

1. Scale : At any given time 1 million streams are running.
   10 million users are connected at any give time to these streams
   
2. Can have slight delay in delivery of comment to other users

I provided the below diagram and explaination , I asked GPT as well to validate the answer

---

Gpt Answer

🛠️ Summary & Verdict

• ✅ Your design is well-structured and scalable for real-time chat.
• ✅ Kafka & Cassandra are solid choices for handling high throughput.
• ✅ WebSocket Manager & Redis make it efficient for tracking active users.

⚡ Potential Enhancements: Redis TTL, Cassandra partitioning, WebSocket failover, comment rate limiting, and optional Elasticsearch for searching comments.

🚀 Final Verdict: 8.5/10 – Excellent design with room for minor optimizations!

---

Final Verdict - REJECT

I have given these optimization as well while communicating, but still feel as unlucky day !