Add AI inference stack: safetensors-zig, tokenizers-zig, vllm-zig, faiss-zig

[SMC17]

Adding four single-purpose Zig 0.16 libraries that compose into an end-to-end CPU LLM inference path. All AGPL-3.0, single-binary, libc-only, with reproducible benchmarks in each repo.

• safetensors-zig — HuggingFace safetensors reader (~5x faster than the Rust upstream on Llama-shape fixtures)
• tokenizers-zig — BPE / WordPiece / Unigram with full HF Encoding parity (189 tests + 600-iter property fuzz)
• vllm-zig — TinyLlama forward pass (RoPE + GQA + KV cache + multi-thread SIMD matmul)
• faiss-zig — vector similarity (Flat + HNSW + IVFFlat + IVFPQ)

Composition write-up: https://smc17.github.io/stax-blog/lab/ai-inference-zig-stack.html

Each repo carries pre-1.0 / "honest non-claims" framing in the README — happy to adjust phrasing if the awesome-zig style guide prefers different wording.

[Copilot]

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

Adds additional Zig-related AI/ML infrastructure libraries to the README’s curated list.

Changes:

• Added four new Zig libraries (safetensors, tokenizers, vLLM substrate, FAISS-like vector search) under the Zig section.
• Included brief performance/feature notes and AGPL-3.0 licensing details for each new entry.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[gemini-code-assist]

Code Review

This pull request adds four new Zig libraries (safetensors-zig, tokenizers-zig, vllm-zig, and faiss-zig) to the README.md. The reviewer suggested sorting these newly added list items alphabetically to maintain organization and readability, providing a concrete code suggestion.

[jiacai2050]

Thank you!