So I realized that bulleting my answers is helping me a lot. Do you recommend that I write down all the actions/results? For example, I tend to need to jam pack as much as possible to prove I was independent. I would love some input into this barebones structure I landed on when I tried to use chat gpt to help me summarize my thoughts. This is THE project I would probably need to present to get leveled at mid level properly. Streaming X Challenge 1. Problem → Need to Ingest 1B+ events of X into the UI from 7 different places (the user should see the events flow into the ui in realtime). Users needed both real-time & historical data updates. The journey started by implementing the old architecture end to end as assigned I found these performance problems below by taking on more of the task by elevating it. However, the volume of writing historical events is NOT as fast as we want to be able to read quickly. 2. Debugging & Fix Attempts → Tried multiple optimizations first Scalability fixes (replicas, indexes, optimized queries). Didn’t fully resolve issue → Had to rethink architecture (Manager came into the picture after I presented the detailed metrics to prove it didn't work after trying the solutions above). 3. Solution → Architectural overhaul + rigorous validation Wrote Design Document. Had to model all the edge cases around the data ordering before I had this meeting. Had meeting: Disagreement about payload size (We can get into this another time) Stateful streams introduced subtle stale data risks. The challenge: ensuring correctness, not just performance. Whiteboarded ordering logic to avoid stale rebroadcasts. Used integration tests to rigorously validate correctness. 4. Impact → Faster, cheaper, and now used in another project Reduced system lag, improved user experience when viewing those high volume data streams. (Measured by checking the system metrics and running scripts) Minimal cost increase, optimized architecture. (Inferred by system metrics and cloud budgeting) Became a standard approach, and I’m now reviewing future implementations in other initiatives. (Measured by new design docs using the same pattern I started)

Bulleting and organization in general is great. There's a reason why I talk about it in-depth here: How does Alex write so well? Your organization looks good to me. Personally, I would organize it into STAR format. But at the end of the day, just do whatever format helps you remember the flow and desired presentation better.

Writing stuff down for Behavioral Preparation

So I realized that bulleting my answers is helping me a lot.

Do you recommend that I write down all the actions/results? For example, I tend to need to jam pack as much as possible to prove I was independent.

I would love some input into this barebones structure I landed on when I tried to use chat gpt to help me summarize my thoughts. This is THE project I would probably need to present to get leveled at mid level properly.

Streaming X Challenge

1. Problem → Need to Ingest 1B+ events of X into the UI from 7 different places (the user should see the events flow into the ui in realtime).

Users needed both real-time & historical data updates.
The journey started by implementing the old architecture end to end as assigned
I found these performance problems below by taking on more of the task by elevating it.
However, the volume of writing historical events is NOT as fast as we want to be able to read quickly.

2. Debugging & Fix Attempts → Tried multiple optimizations first

Scalability fixes (replicas, indexes, optimized queries).
Didn’t fully resolve issue → Had to rethink architecture (Manager came into the picture after I presented the detailed metrics to prove it didn't work after trying the solutions above).

3. Solution → Architectural overhaul + rigorous validation

Wrote Design Document. Had to model all the edge cases around the data ordering before I had this meeting.
Had meeting: Disagreement about payload size (We can get into this another time)
Stateful streams introduced subtle stale data risks.
The challenge: ensuring correctness, not just performance.
Whiteboarded ordering logic to avoid stale rebroadcasts.
Used integration tests to rigorously validate correctness.

4. Impact → Faster, cheaper, and now used in another project

Reduced system lag, improved user experience when viewing those high volume data streams. (Measured by checking the system metrics and running scripts)
Minimal cost increase, optimized architecture. (Inferred by system metrics and cloud budgeting)
Became a standard approach, and I’m now reviewing future implementations in other initiatives. (Measured by new design docs using the same pattern I started)

Writing stuff down for Behavioral Preparation

Discussion