Self-Directed Learning | Local AI workflow setup and optimization for image generation (ComfyUI/Ollama)
Self-directed work installing and configuring local image generation AI workflows. Focused on preparing and running SDXL, Flux, Qwen, and LoRA-based pipelines for generating outputs using node-based systems. The work included validating compatibility and tuning runtime settings for stable generation on NVIDIA hardware. • Installed and configured ComfyUI and Ollama for local inference • Deployed and ran SDXL/Flux/Qwen and LoRA-based model workflows • Troubleshot CUDA, PyTorch, VRAM, and dependency issues affecting generation • Optimized generation settings for low-VRAM environments and verified performance