- architecture teardown

• What it is: one line.
• Scale / constraints: the numbers that shape the design (QPS, data size, latency SLO, geo).
• Sources: engineering blogs, papers, talks (linked).

The problem

What's genuinely hard here? What's the dominant constraint that everything else bends around?

Architecture

The shape of the system - components, data flow, storage. Reference a diagram in assets/diagrams/ rather than inlining a wall of ASCII.

Key decisions

The 3-5 choices that define this system and why they made them. Link to the relevant algorithm or pattern note.

Failure modes & trade-offs

What they gave up, where it bites, how they mitigate it.

Steal / Avoid

• Steal: what generalizes - ideas worth reaching for in your own designs.
• Avoid: what's specific to their scale/context and would be cargo-culting elsewhere.