Applied Science Manager - Perception, Amazon Robotics

About the role

The Vulcan Stow team at Amazon Robotics builds high‑performance, real‑time robotic systems that can perceive, learn, and act intelligently alongside humans at Amazon scale. The team invents and deploys machine learning, optimization algorithms, and geometric reasoning models that enable robots to identify placement opportunities, learn effective stow strategies, and continuously improve capacity utilization. The mission is to deliver robust, real‑time 3D scene understanding that drives downstream manipulation decisions, including semantic occupancy prediction, multi‑view 3D reconstruction, depth estimation, and panoptic segmentation. Amazon is seeking an experienced Applied Science Manager to lead the Perception team, driving technical vision with transformer‑based architectures, generative 3D models, and scalable data pipelines to deliver sub‑250 ms perception outputs for production robots.

Responsibilities

• People Leadership: Motivate, reward, and coach a diverse team of applied scientists and engineers; recruit and retain top talent in 3D computer vision, deep learning, and perception systems; manage day‑to‑day performance and development.
• Technical Vision: Set a vision and create technical roadmaps focused on 3D scene understanding, semantic occupancy prediction, depth estimation, and multi‑view fusion; guide research, design, deployment, and evaluation of perception models such as encoder‑decoder networks, query‑based transformers, and generative architectures.
• Cross‑functional Collaboration: Partner with match & affordances, motion planning, hardware, and fulfillment teams to deliver integrated perception solutions robust to occlusion, clutter, and sensor noise; work with ML infrastructure teams on scalable training pipelines, pseudo‑ground‑truth data generation, and edge deployment optimization.
• Delivery Excellence: Implement best practices in applied research and software development; manage project timelines, resources, and deliverables; contribute to architecture decisions, model reviews, and inference optimization discussions.
• Problem Solving: Participate in deep‑dive troubleshooting and post‑mortem analyses to identify root causes of perception regressions, model failures, depth/segmentation misalignments, and latency degradations.

A day in the life

You will spend a significant portion of your time being a great people manager—motivating, rewarding, and coaching your diverse team. You will recruit and retain top talent while excelling in performance management tasks. You will set the team’s technical vision and roadmap, guiding research, design, deployment, and evaluation of complex computer‑vision and machine‑learning algorithms for contact‑rich, cluttered, real‑world manipulation problems. Collaboration with motion, hardware, and software teams will be essential to create integrated robotic solutions. You will also enforce best practices in applied research and software development, managing timelines, resources, and deliverables effectively.

Benefits

• Medical, Dental, and Vision Coverage
• Maternity and Parental Leave Options
• Paid Time Off (PTO)
• 401(k) Plan

Basic Qualifications

• PhD, or Master’s degree with 4+ years of industry or academic research experience
• 6+ years of applied research experience
• 3+ years of management experience for scientists or machine‑learning engineers
• Experience building machine‑learning models or developing algorithms for business applications
• Proficiency with deep‑learning libraries such as PyTorch, TensorFlow, or MXNet
• Research publications in computer vision, deep learning, or machine learning at peer‑reviewed workshops, conferences, or journals
• Knowledge of engineering practices and patterns for the full software/hardware/network development lifecycle, including coding standards, code reviews, source‑control management, build processes, testing, certification, and live‑site operations

Preferred Qualifications

• Experience hiring and growing top talent
• Deep expertise in 3D perception architectures—including encoder‑decoder networks, vision transformers, multi‑view reconstruction, point‑cloud processing, and generative 3D models for scene completion

Pay

Base salary range: $183,800 – $248,700 USD annually (Seattle, WA).