chore: sync content to repo (#10417)

Co-authored-by: nilbuild <4921183+nilbuild@users.noreply.github.com>

roadmaps/inference-engineering/content/aiperf@zffK61pNUJDwkX7Ukql8y.md

@@ -0,0 +1,10 @@
+# AIPerf
+
+AIPerf is an open-source benchmarking framework designed to evaluate the performance of artificial intelligence hardware and software stacks by measuring their ability to execute large-scale neural network models. It simulates real-world workloads to provide standardized metrics, such as throughput and latency, allowing developers to compare how different systems handle complex computational tasks.
+
+Visit the following resources to learn more:
+
+- [@opensource@aiperf](https://github.com/ai-dynamo/aiperf)
+- [@official@AIPerf Docs](https://docs.nvidia.com/aiperf/welcome-to-ai-perf-documentation)
+- [@official@Profiling with AIPerf](https://docs.nvidia.com/aiperf/getting-started/profiling-with-ai-perf)
+- [@video@AI Perf benchmarking - Dynamo and other LLM endpoints](https://www.youtube.com/watch?v=OuQCSOIR-Y8)

roadmaps/inference-engineering/content/benchmarking--profiling@TxK7B_FF38ms3eKPGug2w.md

@@ -1,3 +1,9 @@
 # Benchmarking and Profiling
 
-Benchmarking and profiling are how inference engineers turn intuition into evidence. Benchmarking measures system-level throughput and latency under load; profiling zooms in to identify which specific operations are consuming time. Both are prerequisites for making optimization decisions with confidence.
+Benchmarking and profiling are how inference engineers turn intuition into evidence. Benchmarking measures system-level throughput and latency under load; profiling zooms in to identify which specific operations are consuming time. Both are prerequisites for making optimization decisions with confidence.
+
+Visit the following resources to learn more:
+
+- [@article@Benchmarking AI Inference Providers: How to Compare Performance Fairly](https://www.gmicloud.ai/en/blog/benchmarking-ai-inference-providers)
+- [@article@Benchmarking AI inference on CPUs: A transparent blueprint for the enterprise](https://next.redhat.com/2026/05/28/benchmarking-ai-inference-on-cpus-a-transparent-blueprint-for-the-enterprise/)
+- [@article@LLM Inference Benchmarking: Fundamental Concepts](https://developer.nvidia.com/blog/llm-benchmarking-fundamental-concepts/)

roadmaps/inference-engineering/content/cuda@DwJJLFLPHGn9zdccTlLY2.md

@@ -1,9 +1,10 @@
 # CUDA
 
-CUDA is NVIDIA's proprietary platform for programming GPUs, consisting of a kernel language (C++), a driver for hardware management, and a runtime API for launching kernels. Every inference engine, framework, and library in the NVIDIA ecosystem depends on CUDA. Understanding CUDA is the foundation for diagnosing performance problems that tools and engines cannot solve automatically.
+CUDA is NVIDIA's proprietary platform for programming GPUs, consisting of a kernel language (C++), a driver for hardware management, and a runtime API for launching kernels. Every inference engine, framework, and library in the NVIDIA ecosystem depends on CUDA, including cuBLAS — NVIDIA's optimized implementation of standard linear algebra routines and the default backend for matrix multiplications in most inference pipelines. Understanding CUDA is the foundation for diagnosing performance problems that tools and engines cannot solve automatically.
 
 Visit the following resources to learn more:
 
-- [@book@CUDA by Example: An Introduction to General-Purpose GPU Programming](https://developer.nvidia.com/cuda-example)
 - [@official@CUDA Programming Guide](https://docs.nvidia.com/cuda/cuda-programming-guide/)
-- [@official@cuBLAS Documentation](https://docs.nvidia.com/cuda/cublas/)
+- [@official@cuBLAS Documentation](https://docs.nvidia.com/cuda/cublas/)
+- [@official@An Even Easier Introduction to CUDA (Updated)](https://developer.nvidia.com/blog/even-easier-introduction-cuda/)
+- [@video@CUDA Programming Course – High-Performance Computing with GPUs](https://www.youtube.com/watch?v=86FAWCzIe_4)

roadmaps/inference-engineering/content/inference-engines@01j1KkEB-nIXxP-rENcJ3.md

@@ -1,3 +1,9 @@
 # Inference Engines
 
-Inference engines are the core serving layer that takes model weights and handles requests at production scale. They implement continuous batching, memory management, parallelism, and optimization techniques so engineers do not have to implement these from scratch. This group covers the three dominant LLM engines and the orchestration platform that sits above them.
+Inference engines are the core serving layer that takes model weights and handles requests at production scale. They implement continuous batching, memory management, parallelism, and optimization techniques so engineers do not have to implement these from scratch. This group covers the three dominant LLM engines and the orchestration platform that sits above them.
+
+Visit the following resources to learn more:
+
+- [@article@What is an Inference Engine & Why it’s Essential for Scalable AI](https://inference.net/content/inference-engine/)
+- [@article@The AI Engineer's Guide to Inference Engines and Frameworks](https://read.theaimerge.com/p/the-ai-engineers-guide-to-inference)
+- [@video@Why so many inference engines..?](https://www.youtube.com/watch?v=_xM8scs4_x4)

roadmaps/inference-engineering/content/kernel-selection--fusion@iKW1wqzd5EfRKvNIRshNN.md

@@ -1,3 +1,10 @@
 # Kernel Selection and Fusion
 
-A CUDA kernel is a function that runs in parallel across thousands of GPU threads simultaneously. Kernel selection is choosing the best existing kernel for a specific operation; kernel fusion combines sequential kernels into one to eliminate unnecessary reads and writes to VRAM between operations. Libraries like cuBLAS, CUTLASS, CuTe, and FlashInfer provide pre-built kernels for common inference operations, and most inference engines perform kernel selection automatically.
+A CUDA kernel is a function that runs in parallel across thousands of GPU threads simultaneously. Kernel selection is choosing the best existing kernel for a specific operation; kernel fusion combines sequential kernels into one to eliminate unnecessary reads and writes to VRAM between operations. Libraries like cuBLAS, CUTLASS, CuTe, and FlashInfer provide pre-built kernels for common inference operations, and most inference engines perform kernel selection automatically.
+
+Visit the following resources to learn more:
+
+- [@article@What Is a CUDA Kernel? A Visual Explainer](https://llm-stats.com/blog/research/what-is-a-cuda-kernel)
+- [@article@From Zero to GPU: A Guide to Building and Scaling Production-Ready CUDA Kernels](https://huggingface.co/blog/kernel-builder)
+- [@video@How to profile CUDA kernels in PyTorch](https://www.youtube.com/watch?v=LuhJEEJQgUM)
+- [@video@Master Cuda Programming: Zero to Hero](https://www.youtube.com/playlist?list=PLVVBQldz3m5v1VDhlCyB1DhPfjREsJWmf)

roadmaps/inference-engineering/content/locust@IQuJFfMP1dW0ELlip6LOd.md

@@ -1,3 +1,9 @@
 # Locust
 
-Locust is a general-purpose open-source load testing framework written in Python. In inference benchmarking it is used to simulate realistic user traffic patterns — including request arrival rates, concurrency ramps, and burst scenarios — that synthetic tools do not capture. It is particularly useful for end-to-end load testing that includes network overhead, queuing, and autoscaling behavior rather than raw model server performance alone.
+Locust is a general-purpose open-source load testing framework written in Python. In inference benchmarking it is used to simulate realistic user traffic patterns — including request arrival rates, concurrency ramps, and burst scenarios — that synthetic tools do not capture. It is particularly useful for end-to-end load testing that includes network overhead, queuing, and autoscaling behavior rather than raw model server performance alone.
+
+Visit the following resources to learn more:
+
+- [@official@Locust](https://locust.cloud/)
+- [@opensource@locust](https://github.com/locustio/locust)
+- [@article@LLM Locust: A Tool for Benchmarking LLM Performance](https://www.truefoundry.com/blog/llm-locust-a-tool-for-benchmarking-llm-performance)

roadmaps/inference-engineering/content/model-formats--runtimes@lTjWXB8YRwx11e8lxSgNm.md

@@ -1,5 +1,8 @@
-# undefined
+# Model Formats and Runtimes
 
-### Model Formats and Runtimes
+Before a model can serve requests, its weights must be stored, loaded, and compiled into an executable form. This group covers the file formats used to distribute weights and the runtimes that compile them into optimized inference engines for specific hardware targets.
 
-Before a model can serve requests, its weights must be stored, loaded, and compiled into an executable form. This group covers the file formats used to distribute weights and the runtimes that compile them into optimized inference engines for specific hardware targets.
+Visit the following resources to learn more:
+
+- [@article@Common AI Model Formats](https://huggingface.co/blog/ngxson/common-ai-model-formats)
+- [@article@Model Weights File Formats in Machine Learning](https://learnopencv.com/model-weights-file-formats-in-machine-learning/)

roadmaps/inference-engineering/content/nvidia-dynamo@ctdMz6rHa4Cwl1x19V_5k.md

@@ -4,4 +4,7 @@ Dynamo is NVIDIA's open-source distributed serving platform that sits above infe
 
 Visit the following resources to learn more:
 
-- [@official@NVIDIA Dynamo Documentation](https://docs.nvidia.com/dynamo/latest/index.html)
+- [@official@NVIDIA Dynamo Documentation](https://docs.nvidia.com/dynamo/latest/index.html)
+- [@official@NVIDIA Dynamo](https://www.nvidia.com/en-us/ai/dynamo/)
+- [@video@Distributed Inference 101: Getting Started with NVIDIA Dynamo](https://www.youtube.com/watch?v=1bRmskFCnqY)
+- [@video@NVIDIA Dynamo in 5 Minutes: What Is It and Why Now?](https://www.youtube.com/watch?v=mXYFcz27eDw)

roadmaps/inference-engineering/content/onnx-runtime@Vmtbdre3SZzwq3FxOO-Bc.md

@@ -4,4 +4,6 @@ ONNX Runtime is an open-source inference runtime that executes models exported t
 
 Visit the following resources to learn more:
 
-- [@official@ONNX Runtime](https://onnxruntime.ai/)
+- [@official@ONNX Runtime](https://onnxruntime.ai/)
+- [@official@ONNX Runtime Tutorials](https://onnxruntime.ai/docs/tutorials/)
+- [@video@ONNX – open format for machine learning models​](https://www.youtube.com/watch?v=0t2jOBZSd6s)

roadmaps/inference-engineering/content/performance-benchmarking@XXfjlqd0CYHwBR5-Kdmmi.md

@@ -1,3 +1,8 @@
 # Performance Benchmarking
 
-Benchmarking measures inference service performance under realistic workloads. Good benchmarks simulate realistic input and output sequence lengths, concurrency patterns, and request contents rather than synthetic uniform inputs. Standard tools include SGLang Genai-bench, NVIDIA GenAI-Perf, and Locust. Always establish a baseline before applying optimizations and change one variable at a time.
+Benchmarking measures inference service performance under realistic workloads. Good benchmarks simulate realistic input and output sequence lengths, concurrency patterns, and request contents rather than synthetic uniform inputs. Standard tools include SGLang Genai-bench, NVIDIA GenAI-Perf, and Locust. Always establish a baseline before applying optimizations and change one variable at a time.
+
+Visit the following resources to learn more:
+
+- [@article@LLM Inference Benchmarking: Fundamental Concepts](https://developer.nvidia.com/blog/llm-benchmarking-fundamental-concepts/)
+- [@article@AI Inference Performance Benchmarks](https://developer.nvidia.com/deep-learning-performance-training-inference/ai-inference)

roadmaps/inference-engineering/content/pytorch@AHVfxQRXmRoiTxC1eeMme.md

@@ -4,4 +4,7 @@ PyTorch is the standard Python framework for building and running neural network
 
 Visit the following resources to learn more:
 
-- [@official@PyTorch Performance Tuning Guide](https://pytorch.org/tutorials/recipes/recipes/tuning_guide.html)
+- [@official@PyTorch Performance Tuning Guide](https://pytorch.org/tutorials/recipes/recipes/tuning_guide.html)
+- [@official@Learn the Basics](https://docs.pytorch.org/tutorials/beginner/basics/intro.html)
+- [@video@Creating a conda Environment for Pytorch and Cuda](https://www.youtube.com/watch?v=_z6WPasIK8U)
+- [@video@How to profile CUDA kernels in PyTorch](https://www.youtube.com/watch?v=LuhJEEJQgUM)

roadmaps/inference-engineering/content/safetensors@jkbrf0_JDljWmVq_o9XE-.md

@@ -1,3 +1,10 @@
 # Safetensors
 
-Safetensors is the standard format for storing and distributing open model weights. It uses memory mapping for fast, zero-copy loading and prevents execution of arbitrary code during deserialization, making it safer than earlier pickle-based formats. All major inference engines load safetensors natively, and most open models on Hugging Face publish weights in this format.
+Safetensors is the standard format for storing and distributing open model weights. It uses memory mapping for fast, zero-copy loading and prevents execution of arbitrary code during deserialization, making it safer than earlier pickle-based formats. All major inference engines load safetensors natively, and most open models on Hugging Face publish weights in this format.
+
+Visit the following resources to learn more:
+
+- [@official@Safetensors](https://huggingface.co/docs/safetensors/index)
+- [@opensource@safetensors](https://github.com/safetensors/safetensors)
+- [@article@SafeTensors: Efficient Serialization Format for Deep Learning](https://medium.com/@nishthakukreti.01/safetensors-efficient-serialization-format-for-deep-learning-57364317be43)
+- [@video@Hugging Face SafeTensors LLMs in Ollama](https://www.youtube.com/watch?v=DSLwboFJJK4&t=2s)

roadmaps/inference-engineering/content/sglang-genai-bench@KHDnY7byKgmaPtC_0uDBM.md

@@ -1,3 +1,8 @@
 # SGLang Genai-bench
 
-SGLang Genai-bench is a benchmarking tool built into the SGLang project for measuring LLM inference throughput and latency under realistic request distributions. It supports variable input and output sequence lengths and concurrent request patterns, making it well-suited for validating SGLang and vLLM deployments before production.
+SGLang Genai-bench is a benchmarking tool built into the SGLang project for measuring LLM inference throughput and latency under realistic request distributions. It supports variable input and output sequence lengths and concurrent request patterns, making it well-suited for validating SGLang and vLLM deployments before production.
+
+Visit the following resources to learn more:
+
+- [@opensource@genai-bench](https://github.com/sgl-project/genai-bench)
+- [@official@GenAI Bench](https://sgl-project.github.io/genai-bench/)

roadmaps/inference-engineering/content/sglang@YtlAf6qloKMmaSr9Bkkl_.md

@@ -4,4 +4,7 @@ SGLang is a high-throughput inference engine with a flexible frontend language f
 
 Visit the following resources to learn more:
 
-- [@opensource@sgl-project/sglang](https://github.com/sgl-project/sglang)
+- [@official@sglang Docs](https://docs.sglang.io/docs/get-started/quickstart)
+- [@opensource@sgl-project/sglang](https://github.com/sgl-project/sglang)
+- [@video@SGLang vs vLLM: Which LLM Inference Framework Should You Use?](https://www.youtube.com/watch?v=18AkpuX3vkY)
+- [@video@Open-source inference with vLLM + SGLang](https://www.youtube.com/watch?v=R74kx7GKEbs)

roadmaps/inference-engineering/content/tensorrt-llm@n2_zIGkQgyYCMYPypz3vb.md

@@ -4,4 +4,7 @@ TensorRT-LLM is NVIDIA's inference engine combining PyTorch-based model executio
 
 Visit the following resources to learn more:
 
-- [@opensource@NVIDIA/TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM)
+- [@opensource@NVIDIA/TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM)
+- [@official@TensorRT-LLM](https://developer.nvidia.com/tensorrt#inference)
+- [@video@TensorRT-LLM is Game Changer](https://www.youtube.com/watch?v=FWhK4O1RXxs)
+- [@video@TensorRT LLM 1.0 Livestream: New Easy-To-Use Pythonic Runtime](https://www.youtube.com/watch?v=sTpQQSbta1k)

roadmaps/inference-engineering/content/tensorrt@M4qjJSEAnsjrmP3A_O8HC.md

@@ -4,4 +4,6 @@ TensorRT is NVIDIA's high-performance inference runtime that compiles PyTorch or
 
 Visit the following resources to learn more:
 
-- [@official@TensorRT Documentation](https://docs.nvidia.com/deeplearning/tensorrt/)
+- [@official@TensorRT Documentation](https://docs.nvidia.com/deeplearning/tensorrt/)
+- [@opensource@TensorRT](https://github.com/nvidia/tensorrt)
+- [@video@Inference Optimization with NVIDIA TensorRT](https://www.youtube.com/watch?v=UnIuMXGylfY&list=PLDvBZlLoGspySJggUWsdB40SyGjIcxaxp)

roadmaps/inference-engineering/content/transformers--diffusers@H_e3KZ7p7enUCYE5M5Lr7.md

@@ -4,6 +4,7 @@ The transformers and diffusers libraries provide reference implementations of po
 
 Visit the following resources to learn more:
 
-- [@course@Hugging Face Course](https://www.youtube.com/playlist?list=PLo2EIpI_JMQvWfQndUesu0nPBAtZ9gP1o)
 - [@official@Hugging Face Diffusers](https://huggingface.co/docs/diffusers/index)
-- [@official@Hugging Face Transformers](https://huggingface.co/docs/transformers/index)
+- [@official@Hugging Face Transformers](https://huggingface.co/docs/transformers/index)
+- [@course@Hugging Face Course](https://www.youtube.com/playlist?list=PLo2EIpI_JMQvWfQndUesu0nPBAtZ9gP1o)
+- [@video@Getting Started With Hugging Face in 15 Minutes](https://www.youtube.com/watch?v=QEaBAZQCtwE)

roadmaps/inference-engineering/content/vllm@wcAtbab81IJdpFh6L-heP.md

@@ -4,4 +4,7 @@ vLLM is the most widely adopted LLM inference engine, offering broad support for
 
 Visit the following resources to learn more:
 
-- [@opensource@vllm-project/vllm](https://github.com/vllm-project/vllm)
+- [@official@vLLM Docs](https://docs.vllm.ai/en/latest/getting_started/quickstart/)
+- [@opensource@vllm-project/vllm](https://github.com/vllm-project/vllm)
+- [@video@vLLM Course](https://www.youtube.com/playlist?list=PLUBL8w2KtS3M2if8k6-Om_7_sIV_swKRA)
+- [@video@What is vLLM? Efficient AI Inference for Large Language Models](https://www.youtube.com/watch?v=McLdlg5Gc9s)

roadmaps/inference-engineering/content/writing-custom-kernels@1dWPVaSQ4TfVU3Obx3m19.md

@@ -5,8 +5,6 @@ Custom kernels are written when existing libraries do not fully utilize a GPU's
 Visit the following resources to learn more:
 
 - [@book@CUDA by Example: An Introduction to General-Purpose GPU Programming](https://developer.nvidia.com/cuda-example)
-- [@official@CUDA Programming Guide](https://docs.nvidia.com/cuda/cuda-programming-guide/)
 - [@official@cuBLAS Documentation](https://docs.nvidia.com/cuda/cublas/)
-- [@opensource@NVIDIA/cutlass](https://github.com/NVIDIA/cutlass)
-- [@opensource@deepseek-ai/DeepGEMM](https://github.com/deepseek-ai/DeepGEMM)
-- [@article@FlashInfer: Efficient and Customizable Attention Engine for LLM Inference Serving](https://arxiv.org/abs/2501.01005)
+- [@official@Introduction to CUDA](https://docs.nvidia.com/cuda/cuda-programming-guide/part1.html)
+- [@opensource@NVIDIA/cutlass](https://github.com/NVIDIA/cutlass)