Haofan Wang is a lead researcher and co-founder with a decade of experience building generative and agentic AI systems for design-centric products and platforms. He founded InstantX, an open-source generative models team sponsored by Hugging Face and fal.ai, and now leads applied research at Lovart AI, shipping advanced in-house models for designers. His work spans industry research roles at Xiaohongshu and Kuaishou and academic stints at CMU, KAUST, and UC Berkeley, blending production-scale engineering with deep learning research. An active open-source contributor, he has meaningfully extended Hugging Face’s popular diffusers library with LoRA integration and SDXL training improvements, helping bring efficient fine-tuning techniques to wider use. Based in Beijing, he’s pursuing a PhD opportunity while remaining open to AIGC business collaborations, combining entrepreneurial drive with hands-on model-building. Colleagues describe him as a practical innovator who moves state-of-the-art research into usable tools for creators.
🤗 Diffusers: State-of-the-art diffusion models for image, video, and audio generation in PyTorch.
Role in this project:
ML Engineer
Contributions:4 reviews, 3 commits, 33 PRs in 1 month
Contributions summary:Frank primarily contributed to the fine-tuning and integration of LoRA (Low-Rank Adaptation) models within the Hugging Face Diffusers framework. They worked on supporting LoRA for the text encoder, converting LoRA checkpoints, and integrating noise offset features. Their contributions also include updates to training scripts for text-to-image generation using LoRA and SDXL, along with bug fixes in SDXL training scripts.
Transfer the ControlNet with any basemodel in diffusers🔥
Contributions:1 review, 44 commits, 1 PR in 18 days
diffuserscontrolnetaigcpytorch
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.