Member of Technical Staff - Image/Video Generation (AI)

TL;DR

Member of Technical Staff (AI): Training large-scale diffusion models for image and video generation with an accent on rigorous ablation studies and architectural optimization. Focus on solving complex distributed training challenges, evaluating generative model performance at scale, and isolating key variables to drive research progress.

Location: Freiburg, Germany and San Francisco, USA

Company

hirify.global is a generative AI research organization behind foundational technologies like Stable Diffusion, Latent Diffusion, and FLUX, dedicated to open science and advancing human creativity.

What you will do

• Train large-scale diffusion transformer models for image and video generation tasks.
• Design and execute rigorous ablation studies to isolate the impact of architectural choices and training variables.
• Analyze speed-quality tradeoffs of neural network architectures in production settings.
• Fine-tune foundation models for specialized applications including inpainting, outpainting, and upscaling.
• Debug and optimize distributed training pipelines to ensure research and production stability.
• Communicate research findings to shape long-term model development strategies.

Requirements

• Proven experience training large-scale diffusion models for image and video data.
• Strong proficiency in PyTorch and transformer architectures.
• Deep expertise in distributed training techniques such as FSDP, model parallelism, and low-precision training.
• Understanding of generative model evaluation metrics and their correlation with actual quality.
• Practical experience fine-tuning models for domain-specific applications.
• Strong problem-solving skills for debugging complex training failure modes.

Nice to have

• Experience writing and optimizing forward/backward Triton kernels.
• Proficiency with GPU profiling and debugging tools like Nsight.
• Published research contributing to generative AI development.

Culture & Benefits

• Focus on research excellence and open science values.
• Opportunity to work on foundational AI models used by millions.
• Direct involvement in high-impact experimental design rather than following established recipes.
• Exposure to scaling challenges at the billion-parameter level.