CS Fundamentals for AI Engineers — Roadmap

1. Operating Systems & Concurrency

• Processes vs. threads, the process/thread memory model
• Concurrency vs. parallelism — the distinction that matters for Python specifically
• The GIL, multiprocessing vs. multithreading for ML workloads
• Memory: stack vs. heap, virtual memory, paging
• CPU caches (L1/L2/L3) and cache-friendly data access
• File systems and I/O basics

2. Networking & Distributed Systems

• HTTP request/response, status codes, REST basics
• TCP/IP vs. UDP, sockets, latency vs. bandwidth
• DNS resolution
• Load balancing strategies
• CAP theorem, consensus (Raft/Paxos intuition)
• Why distributed training and distributed serving are genuinely hard: partial failure, network partitions, stragglers

3. APIs, HTTP & Communication Patterns

• HTTP methods, idempotency, status code families
• REST vs. RPC-style vs. GraphQL
• Authentication vs. authorization, API keys, OAuth 2.0, JWTs
• Polling (short/long), webhooks, Server-Sent Events, WebSockets — when to use which
• FastAPI: path operations, Pydantic models, dependency injection, async endpoints

4. Linux, Git & Developer Tooling

• The shell: pipes, redirection, grep/awk/sed, process management
• The Linux filesystem and permissions model
• Git internals: objects, commits, trees, blobs — not just commands
• Compilers vs. interpreters, what "compiled" actually means for Python/PyTorch

5. Software Engineering Practice

• Design patterns relevant to ML systems (not a full GoF tour)
• Testing philosophy: what to test, what not to test
• Profiling: CPU profiling, memory profiling, finding the actual bottleneck
• Debugging methodology: bisection, reproducing, isolating

Next: Mathematics for AI — the math this systems layer runs.