Generative AI using Amazon SageMaker
A collection of end-to-end SageMaker notebooks and config-driven recipes covering generative-AI model training, fine-tuning, inference optimization, MLOps automation, distributed training, RAG, and agent-based applications.
This repository provides hands-on SageMaker AI implementations spanning the generative-AI lifecycle, from foundational concepts to enterprise-scale deployment. Its model-customization recipes let a user pick one of more than 20 pre-configured foundation models, including Llama, Qwen, DeepSeek, Gemma, Phi, and GPT-OSS, choose a fine-tuning strategy such as QLoRA, Spectrum, or full fine-tuning, and launch a SageMaker training job, with the recipe generator handling dataset formatting, adapter merging, and evaluation automatically.
Beyond the config-driven recipes, deeper hands-on notebooks cover distributed training strategies such as DDP, FSDP, and DeepSpeed ZeRO-3, reinforcement-learning methods such as DPO and GRPO using several specialized frameworks, and diffusion-model fine-tuning such as DreamBooth LoRA for image generation.
Additional material covers retrieval-augmented generation, intelligent agents, and real-world industry applications built on SageMaker.