Software Engineer - Prediction and Behavior ML

Applied Intuition Applied Intuition · Robotics · Sunnyvale, CA · SDS Software Engineering

Software Engineer role focused on developing ML-first behavior prediction and planning modules for autonomous systems, forecasting road user motions and interactions. The role involves prototyping, evaluating, and deploying ML algorithms, leveraging existing infrastructure, and collaborating with perception and planning teams. Requires experience in end-to-end deep learning model development and production software practices.

What you'd actually do

  1. Prototype, evaluate, refine, and deploy state-of-the-art ML algorithms that predict the future motion of the ego agent and other road users in the scene
  2. Leverage established products at Applied Intuition to build the software and infra foundation for behavior-specific ML development
  3. Collaborate tightly with perception and planning engineers on many cross-team initiatives

Skills

Required

  • Experience with the end-to-end development cycle of deep learning models
  • Expertise in subdomains such as modeling, input pipelines, evaluation, deployment, and model optimization
  • 3+ years of experience building production software using modern software practices
  • Fluency in C++, or fluency in Python with intermediate experience in C++

Nice to have

  • Peer-reviewed research at machine learning conferences like NeurIPS, CVPR, ICML, ICLR, ICCV, ECCV, IROS, or ICRA
  • Experience with driver assistance or autonomous driving systems
  • Experience in evaluating and improving system-in-the-loop model performance

What the JD emphasized

  • end-to-end development cycle of deep learning models
  • modeling
  • input pipelines
  • evaluation
  • deployment
  • model optimization

Other signals

  • ML-first behavior prediction and planning
  • forecast the future motions of road users and predict their interactions
  • ML-based planner development
  • state-of-the-art ML algorithms that predict the future motion of the ego agent and other road users