Software Engineer II

Uber Uber · Consumer · New York, NY · Engineering

Software Engineer II role focused on building backend systems for automated retail intelligence, processing image data, orchestrating model inference, and converting predictions into inventory signals. Requires experience with scalable data pipelines, backend service development, and system reliability.

What you'd actually do

  1. Design and maintain robust pipelines that ingest, process, and transform large-scale image and metadata streams into structured inventory signals.
  2. Develop and operate backend services that orchestrate model inference, post-process predictions, and expose high-quality data to downstream systems and partners.
  3. Ensure high availability, low latency, and fault tolerance across distributed systems handling high-throughput image and prediction data.
  4. Design data models and processing logic that translate raw model outputs into accurate, consistent, and interpretable inventory insights.
  5. Partner closely with backend engineers, Machine Learning Engineers, Data Scientists, Mobile Engineers, Product, and Operations teams to define requirements, integrate systems end-to-end, and continuously improve the quality and usability of inventory signals.

Skills

Required

  • B.S. or M.S. in Computer Science, Engineering, or a related field (or equivalent practical experience)
  • 3+ years of professional software engineering experience
  • Strong proficiency in programming languages including C, C++, Java, Python, or Go

Nice to have

  • An understanding of PyTorch or TensorFlow, with a track record of building and training custom architectures.
  • A background in modern CV techniques (e.g., YOLO, Faster R-CNN, Vision Transformers, Siamese Networks) and image processing fundamentals.

What the JD emphasized

  • large volumes of image data
  • orchestrate model inference
  • convert raw predictions into reliable, actionable inventory signals

Other signals

  • orchestrate model inference
  • convert raw predictions into reliable, actionable inventory signals
  • large volumes of image data