Driven Perception Engineer leading advanced 3D sensor fusion and real-time deep learning initiatives for Mitsubishi Electric’s HubPilot project, while pursuing a Master’s in Computer Science and Engineering at the University of Michigan, Ann Arbor.
Automotive Perception Engineer III
Sep 2024 - Present
- Lead the end-to-end development, integration, and testing of HubPilot’s on-board vehicle perception system (HubDrive), surpassing demanding accuracy requirements.
- Research, prototype, and develop advanced 3D perception models, emphasizing sensor fusion approaches.
- Solely designed, developed, deployed, and tested HubPilot’s YardPass feature.
- Plan and manage data collection and labeling pipelines for model training.
- Optimize deep learning models and perception software for NVIDIA embedded SoCs.
- Collaborate with cross-functional engineering teams in Japan and research scientists at MERL (Mitsubishi Electric Research Labs) for various development efforts.
Automotive Perception Engineer II
Jan 2022 - Sep 2024
- Spearheaded the design and development of HubPilot’s HubDrive system from concept to initial deployment, ensuring robust architecture and on-schedule delivery.
- Orchestrated end-to-end data collection and labeling efforts for multiple RGB camera deep learning models, guaranteeing high-quality datasets for model accuracy.
- Supported the deployment of deep learning perception pipelines across various ADAS vision use cases.
- Developed comprehensive flow diagrams detailing the entire perception system design.
- Translated overall system and customer requirements into multi-level perception specifications.
Machine Learning Research Intern
May 2021 - Dec 2021
- Led MEAA’s first in-house High-Performance Compute (HPC) machine build, enabling efficient on-site data processing.
- Partnered with research scientists at MERL to refine model development and deployment processes.
- Developed and optimized a deep learning vision solution for Qualcomm’s SoC (showcased at CES) using the Qualcomm Neural Processing SDK.
- Authored the first end-to-end documentation for MEAA’s deep learning development process, now adopted across R&D teams as a standard reference.