Machine Learning Engineer – Flow Matching & Variational Autoencoders (VAEs)

Lit8 develops generative AI systems for real-time image generation and enhancement. In this role, you will develop, train, and optimize advanced image generative models, focusing on flow matching, rectified flow, and variational autoencoders (VAEs).

We are looking for someone with hands-on experience building text-to-image and/or image-to-image generation systems, with a strong understanding of model training, image quality, controllability, and inference efficiency.

Minimum Qualifications

• At least 2 years of hands-on experience with diffusion, rectified flow, flow matching, or related image generative models.
• Practical experience training, fine-tuning, and evaluating image generation models.
• Hands-on experience with VAEs, latent representations, image encoding/decoding, and reconstruction quality.
• Experience with text-to-image, image-to-image, or both, including conditional generation, editing, enhancement, inpainting, denoising, or controllable generation.
• Understanding of flow-matching objectives, sampling strategies, conditioning, and latent-space generation.
• Strong Python programming skills and proficiency with PyTorch.
• Experience evaluating image quality using perceptual metrics and visual analysis.

Preferred Qualifications

• Experience with large-scale flow-matching models, diffusion transformers (DiTs), or advanced latent generative architectures.
• Experience developing or optimizing high-fidelity VAEs for image generation.
• Experience with distributed training, mixed precision, and large-scale datasets.
• Experience optimizing generative models through distillation, reduced-step sampling, or hardware-aware inference.
• Contributions to open-source generative AI or computer vision projects.

Key Responsibilities

• Develop, train, and improve flow-matching, rectified flow, and VAE-based models.
• Build high-quality image-to-image generation systems.
• Optimize image fidelity, reconstruction quality, controllability, and inference speed.
• Improve training pipelines, sampling methods, and model evaluation.
• Collaborate with research and engineering teams to integrate models into production applications.

What We Offer

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

If you are passionate about flow matching, VAEs, and building high-performance image generative AI systems, we'd love to hear from you.