Mirage
ML Engineer, Generative Video
About this role
Mirage is hiring an ML Engineer to build and optimize large-scale generative video models, focusing on training infrastructure, inference efficiency, and real-time generation capabilities. You'll work at the intersection of research and systems engineering to translate cutting-edge models into production-ready systems.
What you'll do
- Train and optimize large-scale video and multimodal models
- Implement efficiency techniques like distillation, quantization, and pruning for diffusion and autoregressive generation
- Build and maintain distributed training systems with optimized GPU utilization and parallelism
- Develop tooling for experimentation, evaluation, and debugging
- Translate research models into robust, production-ready systems
- Monitor and improve model performance under real-world usage constraints
What they're looking for
- Deep learning systems and infrastructure
- PyTorch and CUDA
- Triton and distributed training (FSDP)
- Model optimization and performance profiling
- Low-latency inference optimization
- Large-scale model scaling and training
- Production systems development
- GPU performance tuning
Benefits
- Comprehensive medical, dental, and vision coverage
- 401K with employer match
- Commuter benefits and meal perks
- Generous PTO policy
- Multiple team offsites and monthly events
- Health and wellness programs
Opens the official application on the employer’s site. No login required.
Mirage
Mirage builds an AI-native video platform that leverages generative media and large language models to enable sophisticated video production, editing, and creative workflows. The company is hiring backend engineers, full-stack software engineers, ML engineers, and iOS developers to advance their AI-driven platform and enhance user experiences in web-based and mobile media creation.
View all jobs at MirageLikely interview questions
- Walk us through your experience optimizing large models for low-latency inference—what techniques did you use and what were the tradeoffs?
- Describe a time you debugged a performance bottleneck in distributed training. How did you identify and resolve it?