Software Development Engineer, Last Mile Delivery, Edge Intelligence

Amazon Amazon · Big Tech · Santa Clara, CA · Software Development

Software Development Engineer role focused on building and optimizing on-device inference pipelines for perception, sensor fusion, and real-time decision-making models on resource-constrained edge devices in Amazon's Last Mile Delivery fleet. The role involves developing applications, optimizing edge compute workloads, and managing device-to-cloud data pipelines.

What you'd actually do

  1. Build and optimize on-device inference pipelines for perception, sensor fusion, and real-time decision-making models, ensuring low-latency and high-reliability performance on resource-constrained hardware
  2. Develop sensor ingestion frameworks and camera/signal processing pipelines that fuse data from multiple modalities (cameras, LiDAR, telematics, GPS) into actionable intelligence
  3. Profile and optimize edge compute workloads for performance, memory, power consumption, and thermal constraints across heterogeneous hardware platforms
  4. Design and implement data pipelines that efficiently move telemetry, video, and sensor data from edge devices to cloud analytics platforms, balancing bandwidth, latency, and cost
  5. Build monitoring, logging, and alerting systems that provide real-time visibility into the health and performance of thousands of deployed edge devices

Skills

Required

  • software development experience
  • design or architecture of new and existing systems
  • large-scale, multi-tiered, multi-threaded, embedded or distributed software applications, tools, systems, and services
  • Object Oriented Design
  • programming with at least one software programming language

Nice to have

  • full software development life cycle, including coding standards, code reviews, source control management, build processes, testing, and operations experience

What the JD emphasized

  • low-latency
  • high-reliability
  • resource-constrained hardware
  • on-device inference

Other signals

  • on-device inference
  • resource-constrained hardware
  • low-latency performance
  • sensor fusion
  • real-time decision-making