The opportunity
Waymo is an autonomous driving technology company with the mission to be the world's most trusted driver. Since its start as the Google Self-Driving Car Project in 2009, Waymo has focused on building the Waymo Driver—The World's Most Experienced Driver™—to improve access to…
What you'll do
Design, implement, and benchmark parameter-efficient fine-tuning (LoRA /: QLoRA) modules in JAX/Flax for multi-task Vision Transformer backbones
Develop dual-level distillation pipelines (intermediate feature matching and: task-head logit distillation) to mitigate multi-task regressions during large-scale data scaling
Collaborate with model optimization, quantization, and latency teams to: validate static weight folding and low-precision quantization, ensuring zero latency overhead on onboard compute platforms
Conduct extensive empirical ablations and evaluate perception metrics on: large-scale autonomous driving datasets across diverse geographic domains
Currently pursuing a PhD or Master's in Computer Science, Electrical: Engineering, Machine Learning, Robotics, or a related technical field
Strong software engineering and deep learning development skills in Python: and modern frameworks (JAX, Flax, PyTorch, or TensorFlow)
What they're looking for
- Solid theoretical understanding and hands-on experience with deep learning: foundation models, Transformer architectures, and multi-task learning
- Experience with model compression, parameter-efficient fine-tuning (e.g.,: LoRA, QLoRA, adapters), or quantization and knowledge distillation techniques
- Publication record at top-tier computer vision or machine learning: conferences (e.g., CVPR, ICCV, ECCV, NeurIPS, ICLR)
- Hands-on experience with model quantization (PTQ, QAT, INT8/INT4/MX4),: low-precision numerics, or hardware-aware model optimization
