Miju Labs

← All buyers · Krea

Erwann Millon

Founding Engineer / ML Researcher · Krea

medium confidenceTier A · Talk to firstImage, video & audio modelsimage

Millon is a founding engineer at Krea and co-author with Sangwu Lee of the FLUX.1 Krea post-training write-up, plus sole author of the Krea Realtime 14B release. He is the most publicly reachable person on Krea's model team — active GitHub, active X, LinkedIn — and the practical entry point to Krea's research side.

Background

Millon came out of the open-source generative-art community (VQGAN+CLIP-era work is associated with his GitHub) into Krea as a founding engineer. He co-wrote the FLUX.1 Krea release in July 2025, describing the SFT-then-RLHF pipeline and the TPO preference-optimisation variant trained on rigorously filtered internal preference data.

In October 2025 he authored Krea Realtime 14B, a 14-billion-parameter real-time long-form video model distilled from Wan 2.1 14B, hitting 11fps in four inference steps on a single B200, with streaming prompt changes mid-generation. That post evaluates quality qualitatively — comparing checkpoints by eye — which is a candid admission of where formal evaluation is missing.

What they run now

Krea's model engineering: post-training and preference optimisation on the image side, and real-time video generation and distillation. The Krea Realtime work is the newest surface — long-form, streaming, interactive video — and it is exactly the kind of model whose quality regressions are invisible to automatic metrics and obvious to a trained eye.

What they buy

Inference, constrained by the same in-house stance as Sangwu Lee. The genuine gap visible in his own writing is the video side: Krea Realtime's quality claims rest on the authors' own eyes, with no rater panel and no rubric. A small, fast, expert panel scoring real-time video checkpoints on motion dynamism, high-frequency detail and temporal artifacts is a product-shaped hole he has publicly described.

Career

  1. c. 2023 – presentFounding Engineer / ML Researcher, Krea

On the record

internal preference data · 2025-07

Co-author of the claim that Krea's RLHF used high-quality internal preference data, rigorously filtered, from labellers who knew the model's weaknesses. source

qualitative judgement · 2025-10

Assesses real-time video checkpoints by observing motion dynamism and high-frequency detail rather than by formal human study. source

Interviews & talks

Why it matters here

High as an access point and moderate as a buyer. He is the reachable half of Krea's model team, and the Realtime video line is where Krea's in-house-only position is weakest — you cannot staff a standing video panel with a handful of engineers when you are shipping checkpoints weekly.

How to reach

Speculation: X or GitHub, short and specific. Open on the Realtime 14B post's own admission that checkpoint comparison is qualitative, and offer a same-week turnaround panel — six named video professionals, a three-axis rubric taken from his own wording, agreement statistics — run once for free on his next checkpoint. Concrete, small, immediately usable, and it sidesteps the 'we do preference data in-house' objection because it is evaluation, not training data.

What we could not establish
  • Whether the Scholar profile under this name is his.
  • His education and pre-Krea history — not found.
  • Whether Krea has any budget line for external services at all.

Why they are on the map

Second author of the Krea 2 report and co-author of the FLUX.1 Krea aesthetics/preference post. source — evidence dated 2026-06

Needs the data for a model; writes the spec.

How Krea buys data

Krea 2 tech report (2026-06-23) says preference annotations were 'collected entirely in house, by people familiar with the specific strengths, weaknesses, and quirks of the model', and that they prototyped reward models with Thinking Machines' Tinker. FLUX.1 Krea post (2025-07) argues that generic preference datasets hurt aesthetics and that under 1M high-quality examples is enough. May 2026: a Contra Labs study using working creatives benchmarked Krea 2 Large on style fidelity. 2026 'ML Researcher - Posttraining' role will 'coordinate with data teams and partners to manage collection of preference data' and 'train custom VLM as reward models'.

Small team (~17 in April 2025, grown since). Founders lead. Research covers Diffusion and Posttraining (RL, reward VLMs, evals). There is an internal data team that does annotation, and a 'Personalization & Recommendation' ML role models user taste.

Known vendors: contra-labs, Thinking Machines Tinker (infra, not data)

Sources

  1. Releasing Open Weights for FLUX.1 Krea (Sangwu Lee & Erwann Millon, 2025-07-31)
  2. Krea Realtime 14B
  3. Erwann Millon — LinkedIn
  4. Krea 2 Technical Report