The library

What I read, and why.

The high-quality resources behind this journey. Papers, docs, and books I actually read, not a link dump. Curated from the field's best.

Architecture

paper

Attention is All You Need

The paper that started it all. Transformer architecture, self-attention, and why it beat RNNs. Still the foundation of every model I serve.

arxiv.org/abs/1706.03762
paper

FlashAttention: Fast and Memory-Efficient Exact Attention with IO Awareness

The kernel that changed inference. IO-aware tiling, online softmax, and why memory traffic beats FLOPs. I built a simplified version of it.

arxiv.org/abs/2205.14135
paper

RoFormer: Rotary Position Embedding

RoPE is how modern models know token order. It's in every model I serve, and it's the reason long context works.

arxiv.org/abs/2104.09864

Inference Optimization Research

paper

Efficient Memory Management for LLM Serving with PagedAttention

The vLLM paper. Virtual memory for the KV cache, near-zero fragmentation, and 2-4x throughput. The reason I use vLLM today.

arxiv.org/abs/2309.06180
paper

Fast Inference from Transformers via Speculative Decoding

The original speculative decoding paper. Draft and target, acceptance sampling, and the math I simulated.

arxiv.org/abs/2211.17192
paper

GPTQ: Accurate Post-Training Quantization

Round-to-nearest with Hessian weighting. The quantization algorithm I built, and the reason 4-bit models work.

arxiv.org/abs/2210.17323
paper

EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty

Feature-level drafting. The successor to plain speculative decoding, and what I covered in the advanced phase.

arxiv.org/abs/2401.15077
paper

Beyond the Buzz: A Pragmatic Take on Inference Disaggregation

The honest look at splitting prefill and decode. When it helps, when it doesn't, and the real tradeoffs. From 2025.

arxiv.org/abs/2506.05508

Developer Tools

tool

vLLM

The serving engine I use daily. PagedAttention, continuous batching, prefix caching. The default choice for a reason.

github.com/vllm-project/vllm
tool

SGLang

RadixAttention and structured output. The engine that made prefix caching a first-class feature.

github.com/sgl-project/sglang
tool

NVIDIA Dynamo

Disaggregated serving made practical. Prefill and decode on separate machines, orchestrated by a scheduler.

docs.nvidia.com/dynamo
tool

LMCache

KV cache management as a service. The layer that makes prefix caching work across replicas.

github.com/LMCache/LMCache

GPU Infrastructure

book

Programming Massively Parallel Processors

The CUDA bible. From thread blocks to memory coalescing. If you want to understand why kernels are fast, this is where you start.

goodreads
book

Site Reliability Engineering

Google's book on running production services. The SLOs, error budgets, and postmortems that apply directly to inference.

sre.google/books
doc

GPU Glossary

Modal's excellent glossary of GPU terms. The reference I keep open when the acronyms get thick.

modal.com/gpu-glossary
doc

NVIDIA Blackwell Architecture Technical Brief

What the next generation of GPUs actually does. FP4, NVLink 5, and the reasoning-era design choices.

resources.nvidia.com
paper

AWQ: Activation-aware Weight Quantization

Only ~1% of weight channels are salient, and they're found by looking at activations, not weights. Scale them up before rounding and you keep uniform INT4. The default in vLLM and TensorRT-LLM.

arxiv.org/abs/2306.00978
paper

Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads

The draft model was inside the target all along. Multiple MLP heads, tree-based attention, verified in a single pass. 2.2-3.6x speedup, no second model to deploy.

arxiv.org/abs/2401.10774
paper

SGLang: Efficient Execution of Structured Language Model Programs

The RadixAttention paper. A radix tree over the KV cache, co-designed with a structured generation DSL. Up to 5x on multi-call programs.

arxiv.org/abs/2312.07104
doc

TensorRT-LLM: In-flight batching with chunked context

Three queues, chunked prefill interleaved with decode, and the whole transformer compiled into fused kernels. Scheduling outside the graph, speed inside it.

nvidia.github.io/TensorRT-LLM

Observability

spec

OpenTelemetry GenAI Semantic Conventions

The emerging standard for LLM telemetry. gen_ai.request.model, gen_ai.usage.input_tokens, and first-class TTFT and time-per-output-token histograms. Instrument once, export anywhere.

opentelemetry.io/docs/specs/semconv/gen-ai
doc

vLLM Production Metrics

The /metrics endpoint that powers real dashboards. gpu_cache_usage_perc, num_preemptions_total, request_queue_size. Alert on preemptions, correlate cache pressure with TTFT spikes.

docs.vllm.ai/en/latest/serving/metrics
tool

Langfuse: trace-level LLM observability

Infrastructure metrics tell you that latency is high. Langfuse tells you why, which prompts are slow, which model regressed, and which users are affected. Cost per trace, quality scores, open source.

langfuse.com/docs/analytics

Production Systems

blog

Baseten Delivery Network: fast cold starts for big models

Cut cold starts 2-3x for models with billions or trillions of parameters, even when burst scaling to 1,000s of replicas. The cold-start problem, solved at the network layer.

baseten.co/blog
blog

Baseten multi-cloud capacity management (MCM)

Unify GPU capacity across clouds, run inference in ours, yours, or both. 99.99% uptime, low latency, compliance-ready. The multi-cloud GPU strategy in practice.

baseten.co/blog
blog

Uber: Scaling AI/ML Infrastructure

How one of the world's biggest real-time systems scales AI/ML infrastructure. The reliability lessons that apply to inference at scale.

uber.com/blog

From the labs (2026)

The inference-engineering posts actually worth reading from the teams building the engines. Collected May-Aug 2026.

blog

Pushing the Limits of Serving DeepSeek-V4-Pro

SGLang on a 1.6T-parameter MoE with FP8 and FP4 weights. The masterclass in serving a frontier model at extreme scale, with the Blackwell optimizations that make it work.

lmsys.org · Aug 19, 2026
blog

Advanced CUDA Graph Techniques in SGLang

Graphing as much of the workload as possible to kill kernel-launch overhead, without breaking compatibility. Breakable CUDA graphs and their production tradeoffs.

lmsys.org · Aug 17, 2026
blog

Unified Radix Cache: One Tree for Hybrid Model Prefix Caching

A single radix tree managing KV cache reuse across full attention and MLA/sliding-window heads. The cache structure for hybrid-attention models like Qwen3.8.

lmsys.org · Aug 11, 2026
blog

SpecForge v0.3.0: Unified Disaggregated & Colocated Speculative Decoding

Speculative decoding training framework with SpecBundle draft models, now supporting both disaggregated and colocated EAGLE3 deployment. Directly integrated with SGLang.

lmsys.org · Aug 4, 2026
blog

HPC-Ops × SGLang: Attention, Router GEMM, and MoE Kernels from Tencent Hunyuan

Battle-tested production kernels from Tencent's large-scale inference: Dynamic Attention and Fused MoE. Real optimizations now in SGLang.

lmsys.org · Aug 7, 2026
blog

Towards Blackwell-Native 8-bit and 4-bit RL: End-to-End MXFP8 and NVFP4 RL in Miles

First implementation of Blackwell-native MXFP8 and NVFP4 RL. The models it produces are natively quantized for serving, which is why it matters for inference.

lmsys.org · Jul 29, 2026
blog

RadixArk × Google: Bringing Full SGLang Features to TPUs

The full SGLang inference stack now runs on Google TPUs with the same feature set as NVIDIA GPUs. Hardware-flexible serving, in practice.

lmsys.org · Jul 30, 2026
blog

MoonEP: Perfectly Balanced Expert Parallelism via Dynamic Redundant Experts

Moonshot AI's open-source expert parallelism library. Dynamic redundant experts for near-perfect load balancing on MoE inference. The Kimi K2/K3 serving stack.

github.com/MoonshotAI/MoonEP · Jul 2026
blog

DeepSpec: Full-Stack Speculative Decoding from DeepSeek

DeepSeek's complete codebase for training and evaluating speculative decoding algorithms, from draft model training to deployment. Spec-decode as a first-class optimization.

github.com/deepseek-ai/DeepSpec · Jun 2026
blog

FlashKDA: High-Performance Kimi Delta Attention Kernels

Moonshot's custom Delta Attention used in Kimi-K2/K3, with flash-attention-style GPU kernels. More KV-cache-efficient than standard MHA, and the key to their long-context serving.

github.com/MoonshotAI/FlashKDA · 2026
blog

MiniMax H3: Open Model Breaking Boundaries Between Tasks and Modalities

MiniMax's new open multimodal model, released with MSA (their custom attention library, open-sourced June 2026). Attention designed for efficient inference across text, image, and video.

minimax.io · Jul 31, 2026
blog

Full-Stack Performance Optimization of AR+DiT in SGL-Diffusion

Replacing the HuggingFace backend with SRT to accelerate autoregressive + diffusion image models. Dedicated TP for AR, sequence-parallelism for DiT, dynamic batching. SGLang beyond text.

lmsys.org · Aug 5, 2026
blog

SGLang Day-0 Support for Qwen3.8-2.4T-A95B

2.4T total parameters, 95B active per token, hybrid attention (full + sliding window). Day-0 support in SGLang for the largest open model. This is what "ready at launch" looks like.

lmsys.org · Aug 12, 2026
blog

How GPT-5.6 fuses frontier intelligence with frontier efficiency

OpenAI's engineering post: load balancing, kernel optimization, speculative decoding, and workload-specific KV config, all driven by GPT-5.6 Sol in Codex. Sol rewrote production kernels in Triton/Gluon for a 20% cost cut, and designed its own draft model for +15% token efficiency. The agentic harness keeps prompt-cache hit rates high with append-only history.

openai.com · Jul 29, 2026
blog

IndexCache: Accelerating Sparse Attention via Cross-Layer Index Reuse

THUDM/Zhipu's optimization for sparse attention that reuses index patterns across transformer layers. Relevant for GLM-4.5-class models using sparse attention for long-context efficiency.

github.com/THUDM/IndexCache · 2026

Books

book

AI Engineering: Building Applications with Foundation Models

Chip Huyen's book. The breadth-first companion to this journey, from evals to deployment.

O'Reilly
book

Build a Large Language Model (From Scratch)

Sebastian Raschka's hands-on LLM architecture book. The code companion to the deep-implementation phase.

Manning
book

AI Systems Performance Engineering

Chris Fregly's new book on building for performance with GPUs, CUDA, and PyTorch. The performance-focused companion.

O'Reilly