Responsibilities:
Own the research-to-deployment cycle for driving models – from literature review and prototyping through to production integration.
Design, implement, and iterate on real-time predictive models, including vision-language-action (VLA) models.
Collaborate on reasoning systems, contributing to VLA models that handle planning across varied horizons.
Bridge cloud-scale training with edge deployment – work on model compression, quantization, speculative decoding, and efficient inference for embedded automotive platforms.
Evaluate and integrate state-of-the-art techniques from the broader AI research community into our autonomy stack.
Collaborate closely with internal R&D teams to unblock technical challenges, accelerate delivery, and raise the overall technical bar.
Requirements:
Ph.D. in Computer Science, Electrical Engineering, Machine Learning, Robotics, or a related field (an MSc with an exceptional background will also be considered).
Strong publication or deployment track record in one or more of: deep learning, computer vision, generative AI, reinforcement learning, or motion prediction.
Demonstrated ability to go from paper to working implementation – not just theory, but shipped systems.
Strong coding skills in Python; experience with C++ is a plus.
Familiarity with modern ML infrastructure: PyTorch, ONNX, Triton, Dynamo, distributed training, model optimization.
Solid mathematical foundations in probability, optimization, and statistics.
Attributes:
Experience with CUDA or low-level GPU optimization.
Hands-on work with model quantization, distillation, or efficient inference on edge devices.
Background in real-time, safety-critical, or embodied AI systems (robotics, autonomous vehicles, drones, etc.).
Experience with foundation models (Language, Vision, Tabular, VLAs) and their on-device deployment.
Familiarity with driving datasets, simulation environments, or sensor fusion pipelines.
