Senior Perception Algorithm Engineer (World Model)

Key Responsibilities

Lead research and development of World Model and Foundation Model approaches for autonomous‑driving perception, covering scene understanding, temporal modeling, multi‑modal representation learning, and perception‑oriented prediction.
Design model architectures and training pipelines that leverage large‑scale production driving data (camera, LiDAR, radar, and vehicle‑state signals).
Build rigorous evaluation methodologies, including scenario‑based metrics, long‑tail case analysis, generalization assessment, and regression tracking.
Translate frontier research into practical perception capabilities; work with engineering teams to validate feasibility, latency, scalability, and deployment constraints.
Identify high‑value data needs, failure modes, and improvement loops from production data; collaborate with data‑infrastructure teams to enhance dataset quality and training effectiveness.
Provide technical leadership on model direction, research roadmap, experiment design, code quality, and future team‑capability building.

Hands‑on experience with World Models, Foundation Models, large‑scale representation learning, generative models, or end‑to‑end autonomous‑driving models directly applicable to perception or scene understanding.
Strong programming skills in Python or C++; ability to write production‑quality research code, debug complex training pipelines, and contribute to engineering implementations.
Practical experience using large‑scale production or fleet data for model training, evaluation, mining, and iteration (beyond small academic datasets).
Deep understanding of modern perception systems such as BEV perception, multi‑sensor fusion, temporal modeling, 3D detection, tracking, occupancy, or scene‑level representation learning.
Ability to define meaningful evaluation metrics and experimental protocols rather than merely optimizing benchmark scores.
Experience leading technically ambiguous projects from research exploration to validated engineering outcomes.

Experience applying large‑scale models to commercial vehicles, robotaxi, autonomous trucks, robotics, or other safety‑critical embodied AI systems.
Familiarity with deployment constraints: inference latency, memory footprint, quantization, TensorRT or similar acceleration frameworks.
Experience with large‑scale distributed training, data curation, synthetic data generation, simulation, or closed‑loop evaluation.
Publications or open‑source contributions in computer vision, autonomous driving, robotics, machine learning, or foundation‑model research, combined with demonstrable engineering delivery.