Companion resources for Chip Huyen's AI Engineering book: chapter summaries, study notes, prompt examples, case studies, and a few analysis scripts. Focuses on engineering practices for adapting foundation models to production rather than step-by-step code tutorials.
AI Train2024
A 671B-parameter Mixture-of-Experts language model (37B activated) trained on 14.8T tokens with 128K context, FP8-first training, a Multi-Token Prediction module, and Hugging Face weights—focused on efficient MoE training and long-context use cases.
Provides end-to-end PyTorch scripts to download/prepare data, implement a transformer from scratch, train LLMs (13M→billion-scale) and generate text. Emphasizes educational clarity and single‑GPU experiments; useful for researchers or hobbyists, but large-scale training still requires substantial compute and engineering.
AI Audio2025
Generates full songs from lyrics and style prompts by first writing an editable melody-and-chord score and then rendering vocals and accompaniment. Unifies symbolic planning and audio synthesis in one pipeline, enabling readable score inspection, zero‑shot covers, and agent-driven editing.
AI Deploy2025
Provides a hardware plugin that runs vLLM on Huawei Ascend NPUs by mapping vLLM execution and memory management to the Ascend runtime. Key features: support for Transformer/MoE/embedding/multimodal models, official docs, CI-backed release branches and community maintenance.
AI Video2025
Retrieval-augmented generation framework for videos spanning hundreds of hours, runnable on a single RTX 3090. Builds multi-modal knowledge graphs over visual and audio content so you can query and chat across many long videos at once.
AI Audio2025
Zero-shot, single‑reference voice cloning TTS with multilingual support (ZH/EN/JA/ES/AR), fine-grained emotion and duration control, and pronunciation hooks (Pinyin/CMU/Kana); ships model weights, Web UI and production deployment recipes for local or server use.
AI Dataset2025
A curated dataset of ~30,000 CUDA kernels generated by an agentic pipeline, including reference PyTorch implementations, runtime metrics, NCU/Torch/Clang-Tidy profiles, error messages and correctness labels — released under CC-BY-4.0 for model fine-tuning and offline RL/optimization research.
AI Infra2025
High-performance CUDA tensor-core GEMM kernel library for LLM workloads: supports FP8/FP4/BF16, fused Mega MoE and MQA scoring, and runtime JIT-compiled kernels. Targets NVIDIA SM90/SM100 and PyTorch—for teams working on low-level GPU kernel optimization.
AI Infra2025
Provides high-throughput, low-latency GPU communication kernels for Mixture-of-Experts (MoE) and expert-parallel workloads, with NVLink↔RDMA-aware forwarding, FP8/BF16 support, and low-latency RDMA hooks for inference decoding.
AI Train2025
An asynchronous, high-throughput framework for large-scale reinforcement learning and agentic training that scales to 1T+ MoE models and 1000+ GPUs, with native verifiers integration, end-to-end SFT/RL/evals, and Slurm/Kubernetes deployment; requires NVIDIA GPUs.
AI Coding2025
Performs fast static type checking and provides a language server with code navigation, semantic highlighting, and completions for Python. Processes ~1.85M lines/sec and completes IDE rechecks typically under 10ms — intended for responsive editor workflows and large codebases.