Machine Learning Engineer – Diffusion and Rectified Flow Image Generative AI

Description du poste

Lit8 develops generative AI systems for

real-time image generation and enhancement

. In this role, you will develop, train, optimize, and deploy image generative models, with a focus on

diffusion, rectified flow, flow matching, and image-to-image generation

.

We are looking for someone who has built image generation systems in practice — training, fine-tuning, evaluating, and improving models with a strong focus on image quality, controllability, inference speed, and production performance.

Minimum Qualifications

• At least

2 years of hands-on experience with diffusion models, rectified flow, flow matching, or closely related image generative models .

• Strong experience with

image-to-image generation

, including image editing, enhancement, inpainting, denoising, super-resolution, style transfer, or controllable generation.

• Hands-on experience training, fine-tuning, evaluating, and debugging image generation models.
• Strong understanding of diffusion training and sampling, conditioning, guidance, latent-space generation, and image-quality optimization.
• Experience with modern generative architectures such as diffusion models, DiTs, U-Nets, transformer-based image models, or flow-based models.
• Strong Python programming skills and hands-on experience with PyTorch or similar frameworks.
• Understanding of image quality evaluation, including perceptual quality, artifacts, sharpness, realism, consistency, LPIPS, SSIM, FID, or similar metrics.
• Strong problem-solving, analytical, and communication skills.

Preferred Qualifications

• Experience with large-scale model training, distributed training, mixed precision, scalable data pipelines, or dataset curation.
• Experience with controllable generation methods such as ControlNet-style conditioning, adapters, LoRA, guidance mechanisms, or prompt/image conditioning.
• Experience optimizing image generation models for real-time, low-latency, or production inference.
• Experience with model op

Annonce publiée sur LinkedIn et reprise par Intelligence Academy. Seule l'annonce d'origine fait foi ; nous ne sommes pas l'employeur.