Staff Engineer, Perception Systems

Company: Rivian

Location: Palo Alto, California, USA (onsite)

Employment type: Full-time

Overview


Rivian is hiring a Staff Engineer to join its Perception Systems team within the broader perception group, focused on the data-loop, triage and model evaluation processes that sit behind deploying perception models into Autonomy features. The role centres on root-cause analysis of perception system underperformance and building software at the interface between perception and the wider autonomy stack. It’s a systems-level position that calls for first-principles thinking and the ability to identify issues invisible to a single domain specialist. Candidates will work across the perception group and systems engineering to standardise how performance is understood and improved.

Key Responsibilities

  • Lead root-cause analysis of model performance issues from a system-level and sensor-performance lens, including building software and tooling to streamline RCA, data analysis, and evaluation workflows
  • Analyse data-loop processes and triage results to specify requirements for improved data collection, labelling, or augmentation that enhance model robustness
  • Establish, review, and maintain evaluation metrics, developing robust KPIs to identify and prevent regressions in perception models and implementing these in code
  • Collaborate cross-functionally with teams across the perception group and systems engineering to address technical challenges and standardise performance understanding

Requirements


Essential: a Bachelor’s, Master’s or PhD in Computer Science, Robotics, Electrical Engineering, or a related field; demonstrated experience deploying autonomy in the real world, including strong familiarity with machine learning in a production environment; strong problem-solving and coordination skills with the ability to operate effectively in a highly cross-functional, fast-paced environment; strong cross-functional communication and coordination skills.
Desirable: none specified in the source listing.