What should every engineer know about concurrency and operating systems?

Concurrency is structuring a program to handle many tasks at once; parallelism is executing them simultaneously on multiple cores.

Essentials

• Processes vs threads — processes are isolated; threads share memory (faster, but riskier)
• Async / event loop models (Node.js, Python asyncio) handle many I/O-bound tasks on few threads; CPU-bound work needs worker threads or processes
• Race conditions and shared state — protect with locks, atomic operations, immutable data or message passing
• Deadlock — avoid by acquiring locks in a consistent order and using timeouts
• Memory — stack vs heap, garbage collection, leaks and pressure; profile before tuning
• OS resources — file descriptors, sockets, scheduler behaviour and context-switch cost; ulimits and pool sizes matter under load
• Backpressure — bound queues and concurrency so overload degrades gracefully

Diagnosis tools: profilers, flame graphs, thread dumps, strace/perf and heap snapshots.