About the role
What we're looking for?
We are seeking a Machine Learning Engineer with at least 2 years of full-time experience in production ML systems. You should be comfortable working with GPUs, debugging CUDA issues, and have a track record of optimizing large models for performance. Bonus points if you have experience with video or audio models and low-level optimization techniques like CUDA kernels.
What you'll do:
- Take ownership of model performance, profiling and improving inference for latency and throughput using techniques like quantization, pruning, and distillation.
- Work closely with the research and infrastructure teams to design and implement efficient data pipelines for video data at a petabyte scale.
- Apply model acceleration techniques (e.g., TensorRT, ONNX, vLLM) to optimize multimodal models including video diffusion, LLMs, and speech models.
- Lead the development of evaluation frameworks to measure model quality and guide continuous improvement.