• Advance core computer vision model performance (object detection, segmentation, OCR) for warehouse inventory scanning across drone and MHE Vision platforms • Own the full ML lifecycle from research and experiment design through production deployment and monitoring — applying rigorous ablation studies and SOTA methodology • Collaborate with the ML infrastructure team on model optimization and deployment across cloud and edge inference targets (ONNX, TensorRT, quantization) • Work with Operations and Product to understand customer needs and translate them into ML improvements with measurable business impact • Provide technical leadership and mentorship to the ML team, raising standards for experiment design, model evaluation, and production readiness • Explore next-generation perception capabilities, including embedded and on-prem inference optimization for new deployment targets

• Develop learning-based manipulation models for end to end sensor-driven interaction (e.g., reaching, motion generation, and execution in dynamic environments). • Build and maintain manipulation training pipelines: dataset creation from robot logs/teleop, action representations, augmentation, and distributed training. • Design evaluation metrics and regression tests that quantify manipulation reliability, recovery behavior, and safety in real environments. • Develop sim-to-real workflows for manipulation learning, including simulation environments, domain randomization, and failure-mode testing. • Optimize and distill models for edge deployment; benchmark latency, memory use, and stability on target hardware. • Partner with the AI platform team to integrate policies with control and safety systems, and validate end-to-end performance on robots. • Analyze field performance, identify dominant failure modes, and drive iterative improvements through data collection and targeted retraining.

• Develop learning-based navigation models that predict safe, smooth trajectories from sensor inputs and/or perception representations. • Build imitation learning pipelines from fleet logs (trajectory extraction, filtering, scenario balancing, evaluation). • Implement simulation-based refinement (RL, reward shaping, domain randomization) to improve robustness. • Define navigation success metrics aligned to product outcomes. • Collaborate with the AI Platform team to integrate learned policies behavior/safety systems and validate on-robot. • Build regression tests and scenario replay suites for challenging scenarios. • Analyze field behavior, identify failure modes, and close the loop through data curation and retraining.

• Develop multimodal world-model architectures that ingest and fuse camera, LiDAR/depth, and robot state and produce short-horizon predictions. • Build and maintain training pipelines: dataset construction, tokenization/backbones, distributed training, and ablation frameworks. • Define model evaluation metrics and regression suites that reflect real robot outcomes. • Create visualization/debug tooling for temporal predictions (rollouts, replays, overlays, failure case inspection). • Optimize and distill models for edge deployment; benchmark latency, memory, and stability on target hardware. • Collaborate with the AI Platform team to integrate the world model into autonomy stacks and validate behavior. • Work with Operations to identify failure modes in the field and drive data curation and model iteration.