Loading...
Loading...
Found 13 Skills
Pick the serving stack and per-runtime memory flags (vLLM, SGLang, llama.cpp, TensorRT Edge-LLM) for an LLM/VLM workload on any NVIDIA Jetson.
End-to-end SGLang SOTA performance workflow. Use when a user names an LLM model and wants SGLang to match or beat the best observed vLLM and TensorRT-LLM serving performance by searching each framework's best deployment command, benchmarking them fairly, profiling SGLang if it is slower, identifying kernel/overlap/fusion bottlenecks, patching SGLang code, and revalidating with real model runs.
Serves AI models on AMD Instinct GPU hardware using vLLM. Use this skill whenever the user wants to run, serve, deploy, start, host, or launch a language model on an AMD GPU, AMD Instinct, MI300X, MI325X, MI350X, or MI355X. Also use when the user mentions vLLM on ROCm, vLLM on AMD, serving on HBM, or asks how to get a model running on AMD data center hardware. Use when the user asks "run Qwen3", "serve DeepSeek", "start a vLLM endpoint", "get a model running on my AMD machine", or any similar phrasing. Handles the full flow: GPU detection, environment validation, vLLM configuration, launch, and health verification. Do not use for NVIDIA GPUs, consumer AMD GPUs (RX series, Radeon), Ryzen AI, NPU, MI250X, or MI100.
Deploy and manage vLLM for high-throughput LLM inference. Configure continuous batching, tensor parallelism, quantization, and OpenAI-compatible API endpoints for production LLM serving.
Benchmark vLLM or OpenAI-compatible serving endpoints using vllm bench serve. Supports multiple datasets (random, sharegpt, sonnet, HF), backends (openai, openai-chat, vllm-pooling, embeddings), throughput/latency testing with request-rate control, and result saving. Use when benchmarking LLM serving performance, measuring TTFT/TPOT, or load testing inference APIs.
Fast structured generation and serving for LLMs with RadixAttention prefix caching. Use for JSON/regex outputs, constrained decoding, agentic workflows with tool calls, or when you need 5× faster inference than vLLM with prefix sharing. Powers 300,000+ GPUs at xAI, AMD, NVIDIA, and LinkedIn.
Develop, debug, and optimize SGLang LLM serving engine. Use when the user mentions SGLang, sglang, srt, sgl-kernel, LLM serving, model inference, KV cache, attention backend, FlashInfer, MLA, MoE routing, speculative decoding, disaggregated serving, TP/PP/EP, radix cache, continuous batching, chunked prefill, CUDA graph, model loading, quantization FP8/GPTQ/AWQ, JIT kernel, triton kernel SGLang, or asks about serving LLMs with SGLang.
Run vLLM performance benchmark using synthetic random data to measure throughput, TTFT (Time to First Token), TPOT (Time per Output Token), and other key performance metrics. Use when the user wants to quickly test vLLM serving performance without downloading external datasets.
Run an autonomous Humanize-governed vLLM SOTA performance loop for one LLM model: first perform the fixed fair vLLM/SGLang/TensorRT-LLM deployment search and benchmark, then start one RLCR loop that repeatedly decides the gap, profiles the current bottleneck, runs layer/kernel pipeline analysis, patches vLLM code, optionally uses ncu-report-skill for kernel evidence, and revalidates until vLLM matches or beats the best observed framework under the same workload and SLA.
Deploys and optimizes AI/ML inference workloads on GKE, using GPUs, TPUs, and model servers. Use when deploying GKE inference servers, configuring GKE GPU resources for inference, or deploying LLMs on GKE. Don't use for generic batch jobs or HPC task queues (use gke-batch-hpc instead).
Run an autonomous Humanize-governed SGLang SOTA performance loop for one LLM model: first perform the fixed fair SGLang/vLLM/TensorRT-LLM deployment search and benchmark, then start one RLCR loop that repeatedly decides the gap, profiles the current bottleneck, runs layer/kernel pipeline analysis, patches SGLang code, optionally uses ncu-report-skill for kernel evidence, and revalidates until SGLang matches or beats the best observed framework under the same workload and SLA.
Use when the user asks about MLX serving, mlx_lm.server, oMLX, Apple Silicon LLM serving, or running a local LLM on a Mac, and when troubleshooting a model that fails to load, OOM during load or inference, a server that hangs or crashes at batch>1, tool calls returned as plaintext content, a throughput regression, or the choice between mlx-lm and oMLX. Also covers oMLX feature-flag tuning (turboquant_kv, dflash, MTP, specprefill, thinking_budget, max-concurrent-requests, force_sampling), the OptiQ proxy for models exceeding RAM, Llama-4 ChunkedKVCache batch handling, Llama-3 tool-call JSON format, and bench-driven validation of serving configs. Apple Silicon M-series only, not cloud LLM hosting, non-MLX backends such as llama.cpp, Ollama and vLLM, or model training.