# Erwann Millon

> founding engineer @ krea.ai

Location: San Francisco, California, United States
Profile: https://flows.cv/erwann

ML researcher / engineer specialized in image/video gen AI
Turning cutting-edge research into powerful tools for creatives

## Work Experience
### Founding Engineer @ krea.ai
Jan 2023 – Present | San Francisco Bay Area
- co-led research on Krea 1 and flux1.krea, SOTA image models used by millions
- published and presented at major conferences (SIGGRAPH, ICML), cited by labs like Nvidia, Tencent, Huawei
- trained Krea Realtime 14B, world's largest open-source realtime video model
- trained style reference system for Krea 1, allowing t2i with style conditioning from reference images
- built world's fastest inference for realtime t2i and image editing with LoRA swapping
- designed versatile finetuning pipeline allowing users to train LoRAs on Flux in <2 minutes
- built and scaled inference and infra on k8s to millions of users

### Machine Learning Engineer @ Unknown
Jan 2021 – Jan 2023
Notable Projects: 

FashionGen
• Built a multi-model Al pipeline (Bounding boxes w/ segmentation models -> Parallelized BLIP image-to-caption -> Stable diffusion finetuning)
• Built an interface allowing users to generate novel outfits
• Indexed CLIP embeddings of fashion images for text
and visual semantic search of fashion items
• Multi-GPU training on AWS EC2

Virtual Clothes Try-On (Dreambooth)
  - Used 10-20 images of individual clothing articles to finetune Stable Diffusion allowing users to apply a real outfit onto a picture of themselves.
  - Built automated pipeline to scrape images from a product page, process them, and launch distributed training
  - Used textual-inversion and prompt engineering to optimize results

Latent Face Editor
  - Built full Web UI for face editing
  - Implemented adversarial gradient-masked loss to allow localized edits. 
  - Wrote custom backpropagation hooks in the PyTorch autograd engine

Color Diffusion
  - World's first diffusion model to tackle image colorization
  - Built experience training conditional diffusion models from scratch
  - Learned about model pruning, hyperparameter search, hardware-specific optimization

VQGAN-CLIP reimplementation
  - Simple implementation focused on readability and ease-of-use: 3 lines of code to generate or edit an image
  - Designed to be clean, modular, extensible
  - Featured as an official research project on the Huggingface transformers repo

### Translator (French/English) and Voice Over Artist @ Unknown
Jan 2019 – Jan 2022
- Led large-scale multi-month projects
- Specialized in e-learning, clients include the Government of Canada, UNICEF, UPL
- Experience from translation to performing to audio engineering

### Co-Founder @ Careceiver
Jan 2019 – Jan 2021 | Lisbon Area, Portugal
- Startup founded during European Innovation Academy
- Built an MVP for a SaaS that helps caregivers care for chronically ill relatives
- Incubated at HAG startup accelerator


## Education
### Architecte Du Numerique (Accelerated Masters' Equivalent) in Computer Science
42 Paris

### Startup Launchpad Program in Entrepreneurship/Entrepreneurial Studies
European Innovation Academy

### Centre International de Valbonne

### Full Stack Web Development Certification in Computer Software Engineering
freeCodeCamp


## Contact & Social
- LinkedIn: https://linkedin.com/in/erwann-millon
- GitHub: https://github.com/ErwannMillon

---
Source: https://flows.cv/erwann
JSON Resume: https://flows.cv/erwann/resume.json
Last updated: 2026-04-10