Anzhella P.

Image/Video Generation | Python Developer | Generative AI Engineer | ComfyUI Workflows | Krea, Z-Image, SDXL, Flux, Nano Banana Pro, ControlNet I’m a Senior Generative AI Engineer with 6+ years in Deep Learning (Python, model training, optimization) and 4+ years in image & video generation, working with cutting-edge AI startups. I design production-level AI products that combine technical depth with practical use, train image/video generation models (LoRA & full finetuning), and write custom nodes and scripts. 🌍 Highlights & Experience: - Built an AI fashion platform MVP with 50+ digital faces, consistent outputs, and end-to-end workflow for fashion images. - Trained LoRA and finetuned Krea, Z-Image, Qwen, Flux, Stable Diffusion models for art styles, realistic characters, and brand identity use cases. - Developed ComfyUI workflows for image and video generation, including WAN-based pipelines for video2video, face and head swap, and influencer content. - Created automated AI photoshoot systems for fashion, lifestyle, and e-commerce campaigns, reducing time and cost for brands. - Built an AI filmmaking engine to produce consistent characters and storyboard with video animation. ⚡ What I deliver: - Custom model training (LoRA & finetuning) tailored to dataset and brand goals - Production-ready workflows in ComfyUI / Nano Banana / Z-Image / Flux / SDXL / Video Generation - Face / head swap & character pipelines with high consistency and control (ControlNet) - Python-based automation for dataset filtering, captioning, and batch generation - Cloud solutions (Runpod, Replicate, Vast.ai, custom GPU servers) for scalable training & inference - End-to-end delivery: from raw dataset → trained model → working workflow 🛠️ Tools & Stack: Python, PyTorch, ComfyUI, LoRA, ControlNet, IP-Adapter, Kohya SS 🧩 Models I work with: Z-Image, Image Qwen, SDXL, Flux, NanoBanana, Stable Diffusion, WAN 2.2, LTX, HunyuanVideo, VEO 💡 My focus: AI workflows that actually save brands time, money, and energy while delivering visuals people want to share.