Popular repositories Loading

1. qlora-finetuning-and-attention qlora-finetuning-and-attention Public

   QLoRA fine-tuning of two small open LLMs (decoder-only and encoder-decoder) for audience-adaptive Q&A, plus from-scratch scaled dot-product attention in PyTorch.

   Jupyter Notebook 1

2. customer-service-analytics-agent-mohammad-ghanayim customer-service-analytics-agent-mohammad-ghanayim Public

   Python

3. k8s-distributed-llm-finetuning k8s-distributed-llm-finetuning Public

   Multi-node PyTorch DDP fine-tuning of a causal LM on Nebius GPU Kubernetes, with SkyPilot workload orchestration — a 2-node torchrun job with verified NCCL collectives.

   Python

4. observable-vllm-text2sql-agent observable-vllm-text2sql-agent Public

   Text-to-SQL agent on vLLM (Qwen3-30B-A3B) with a LangGraph verify→revise loop, instrumented on two observability planes for metric-grounded SLO diagnosis.

   Python

5. gpu-cuda-inference-optimization gpu-cuda-inference-optimization Public

   Three measured notebooks on GPU inference optimization: roofline analysis, KV-cache decode optimization (4.21x), and CUDA-graph launch-overhead elimination (5.38x). Pure PyTorch.

   Jupyter Notebook