Scott Glover
Community@scottgl9 · Houston, TX
Agent Skills by Scott Glover
Showing 16 vetted skills indexed across 1 GitHub repositories.
scottgl9sglang-diffusion-modelopt-quant
Quantize diffusion transformers to FP8 or NVFP4 with NVIDIA ModelOpt.
scottgl9sglang-diffusion-benchmark-profile
Benchmark and profile diffusion models using torch.profiler traces.
scottgl9sglang-diffusion-add-model
Automates integration of new diffusion models into the SGLang framework.
scottgl9sglang-diffusion-ako4all-kernel
Tune and validate SGLang diffusion kernels with AKO4ALL profiling and benchmarking workflows.
scottgl9sglang-diffusion-performance
Configures SGLang diffusion models for optimal speed and resource efficiency across GPU setups.
scottgl9clean-startup-log
Suppress non-critical warnings and verbose outputs from server startup logs.
scottgl9write-sglang-test
Guide SGLang CI/UT test writing, registration, and automation.
scottgl9ci-workflow-guide
Design, manage, and debug GitHub Actions CI pipelines with gating and retry logic.
scottgl9add-jit-kernel
Integrate custom CUDA JIT kernels into the sglang framework with Python wrappers.
scottgl9debug-cuda-crash
Record CUDA kernel calls and tensor states to diagnose GPU errors.
scottgl9generate-profile
Launch, validate, and profile SGLang server instances with Chrome and Perfetto traces.
scottgl9sglang-auto-benchmark
Automate repeatable LLM benchmarking across hardware and model configurations.
scottgl9add-sgl-kernel
Implement and integrate CUDA/C++ kernels into sgl-kernel for tensor operations.
scottgl9llm-torch-profiler-analysis
Analyzes LLM torch-profiler traces to identify performance bottlenecks and fusion opportunities.
scottgl9sglang-bisect-ci-regression
Diagnose CI test regressions by extracting failure signatures and bisecting commits.
scottgl9debug-distributed-hang
Diagnose hanging issues in multi-GPU distributed inference using py-spy, NCCL logs, and CUDA coredumps.