Description

• Train and fine-tune GenAI models to generate hyperrealistic synthetic media (image and/or video).
• Curate and prepare datasets specifically for generative training (identity coverage, lighting/pose balance, captioning/metadata, filtering, de-duplication).
• Improve realism and likeness by iterating on fine-tuning strategy (adapters/LoRA-style approaches, full fine-tunes where needed), and systematic evaluation.
• Diagnose and solve generative failure modes (identity drift, artifacts, texture issues, uncanny results, overfitting to subjects, prompt/conditioning sensitivity).
• Manage multiple training runs and track progress (prompt/eval sets, side-by-side comparisons, regression checks, experiment logs).
• Critically evaluate results with a focus on facial likeness, realism, consistency, and “production-readiness.”
• Collaborate with R&D to innovate GenAI training techniques and integrate practical improvements into workflows.
• Stay current on emerging GenAI research, tools, and workflows; prototype creative implementations.
• Work with data scientists, software engineers, and product managers to translate creative/production needs into technical solutions.
• Monitor and optimise model performance for efficiency and quality (inference speed, VRAM, throughput, consistency).

• Proven experience with generative model training/fine-tuning (Stable Diffusion-family or similar; diffusion/GAN/transformer-based generative).
• Strong dataset skills for GenAI (curation, balancing, subject coverage, filtering, evaluation sets).
• Familiarity with UNIX/Linux environments and GPU training workflows.
• Critical eye for facial likeness and visual quality; able to judge and explain issues clearly.
• Excellent teamwork and communication skills.
• Strong problem-solving and out-of-the-box thinking.
• Calm under pressure and able to meet deadlines.

• VFX or compositing experience (Nuke or similar), especially facial/beauty integration.
• Python + ML engineering experience (training automation, data pipelines, evaluation tooling).
• Experience with face-swap systems and face workflows.

À propos de nous
Nous sommes DNEG, l'un des principaux studios d'effets visuels et d'animation au monde pour la création de longs métrages primés, de contenu télévisuel et multiplateforme. Nous employons plus de 9000 personnes avec des bureaux et des studios dans le monde entier: en Amérique du Nord (Los Angeles, Montréal, Toronto, Vancouver), en Europe (Londres), en Asie (Bangalore, Mohali, Chennai, Mumbai) et en Australie (Sydney).

Chez DNEG, nous croyons fondamentalement que l'acceptation de nos différences est un élément essentiel de notre réussite collective. Nous nous engageons à créer un environnement de travail équitable, diversifié et inclusif pour nos équipes mondiales, où chacun se sent important et à sa place. Nous accueillons et encourageons les candidatures de tous, quels que soient le parcours, l'expérience ou le handicap. N'hésitez pas à nous faire savoir si vous avez besoin d'ajustements ou de soutien au cours de la procédure de candidature. Nous avons hâte de vous rencontrer!