Static analysis + LLM enhancement. A PoC that found a bug 32x faster than raw AI crawling.

Why AI tools get lost in large codebases, and an idea I'm exploring to fix it.

The two fundamental ways encryption algorithms process data.

The two notification modes for I/O multiplexing, why edge-triggered is trickier, and when each one matters.

Both let processes talk to each other, but the abstraction level changes everything about how you use them.

How to tell where one message ends and another begins, and why getting it wrong opens you up to buffer overflow and DoS attacks.

I don't like Java, I love Zig and Rust, and neither of those feelings matter when network latency is 10,000x bigger than your FFI overhead. A case study in not rewriting things.

I thought Zig would be simpler and Rust harder. After building the same project in both, I learned that language simplicity doesn't matter when the tooling fights you at every step.

Why experience beats knowledge, how humans actually learn, and why you should say yes to shitty projects. A reflection on 10 years of getting humbled.

How I'm learning Rust and Zig simultaneously while avoiding the trap of AI-assisted development. Knowledge bases, project logs, and why suffering is essential to learning.

The two places your data lives, how they differ, and the use-after-free bug that taught me why it matters.

How I learned that Zig's HashMap stores pointers, not data, and why that creates use-after-free bugs if you're not careful.

The two string types in Rust, when to use each, and why the distinction actually matters for memory management.

Rebuilding a p2p e2e secret sharing cli tool from scratch