Machine Learning Engineer – Model Distillation for Real-Time Image Generative AI

Lit8 develops generative AI systems for real-time image generation and enhancement. In this role, you will focus on distilling and optimizing image generative models to achieve high visual quality with significantly lower inference latency and computational cost.

We are looking for someone with proven hands-on experience applying model distillation techniques to image generation, particularly diffusion-based and image-to-image models.

Minimum Qualifications

• At least 2 years of hands-on experience distilling image generation models, preferably image-to-image models.
• Strong practical experience with teacher-student training, progressive distillation, consistency distillation, adversarial distillation, or related techniques.
• Experience working with diffusion, transformer-based, GAN-based, or other image generative architectures.
• Hands-on experience training, fine-tuning, debugging, and evaluating distilled models.
• Experience preserving image quality, controllability, and visual consistency while reducing inference steps or model complexity.
• Strong Python programming skills and proficiency with PyTorch.
• Understanding of image quality evaluation, inference latency, memory efficiency, and perceptual metrics.

Preferred Qualifications

• Experience distilling diffusion or flow-based models into few-step or single-step generators.
• Experience with image-to-image generation, including editing, enhancement, denoising, inpainting, or super-resolution.
• Experience with advanced distillation techniques, including adversarial diffusion distillation or consistency-based approaches.
• Experience optimizing generative models through quantization, pruning, mixed precision, or hardware-aware tuning.
• Experience deploying distilled models in real-time or production applications.
• Contributions to open-source generative AI or model optimization projects.

Key Responsibilities

• Develop and implement model distillation techniques for image generative AI.
• Train and optimize distilled models for real-time or low-latency inference.
• Preserve image quality and controllability while reducing computational requirements.
• Benchmark distillation strategies across model architectures and image generation tasks.
• Collaborate with research and engineering teams to integrate optimized models into production applications.

What We Offer

• The opportunity to work on advanced model distillation and efficient generative AI systems.
• A research-driven environment with real product impact and technical ownership.

If you are passionate about model distillation and making advanced image generation systems faster and more efficient, we'd love to hear from you.