About the Team

The Constraint Generation team is responsible for translating perception and prediction inputs into actionable options for the planner. It is one of the core motion planning teams, whose work directly impacts how our autonomous trucks drive. The team interfaces with perception, mapping, prediction, trajectory generation, trajectory evaluation, and other planning teams in order to implement and improve onboard features. In addition, it works with a wide range of stakeholders throughout the company (systems, simulation, metrics) to deploy these features safely and monitor their performance. This team’s work addresses challenges specific to trucking (high speed highway operations, long stopping distances, complex turn geometries, and more), and is central to the performance of the AV system on the road.

About the Role: 

We are looking for strong software engineers with motion planning experience to help develop and deploy motion planning components for autonomous trucks. In this role, you will work across all stages of the development lifecycle to design, implement, test, and evaluate new features and feature improvements that directly impact the self-driving system’s on-road performance.

A core focus of this role is constraint generation, which defines the solution space for trajectory generation by establishing the boundaries, corridors, and high-level decisions for candidate trajectories. This covers critical real-world trucking maneuvers, including:

• Targeting gaps during merges and lane changes while reasoning about traffic congestion and proactive deceleration.
• Determining optimal locations to pull over safely to the shoulder during degraded or emergency states.
• Constructing dynamic driving corridors through active construction zones where valid HD maps do not exist and navigation relies purely on real-time perception.
• Pre-positioning the vehicle prior to tight, sharp turns to guarantee clean, safe completion.

Our team is small, product-focused, and highly collaborative. You will work closely with perception, prediction, and mapping teams to ensure the planner receives the necessary upstream signals in an optimal format, while delivering high-performance modern C++ that operates reliably on compute-constrained vehicle platforms.

Responsibilities:

• Design & Feature Architecture: Frame ambiguous vehicle behaviors and design new features or targeted behavior improvements (e.g., addressing late-braking behavior during merges in dense traffic, or architecting a robust shoulder pull-over pipeline).
• Onboard Implementation & Testing: Write, integrate, and optimize clean modern C++ on the vehicle stack, validating behavior through rigorous simulation, log replays, and unit testing.
• On-Vehicle Debugging & Diagnostics: Investigate and debug complex edge cases observed during on-track and on-road testing (e.g., diagnosing driver disengagements caused by unintended brake stabs).
• Interface & Systems Collaboration: Negotiate clean architectural interfaces with upstream perception, world fusion, and prediction teams, as well as downstream trajectory generation and evaluation modules.
• Verification & Validation: Partner with systems and verification teams to ensure safety-critical validation criteria are met before deploying code to the autonomous fleet.

Qualifications:

• Strong C++ & Python skills, particularly in resource-constrained environments
• Exceptional written and verbal communications skills
• Experience developing code for real-world robotics/semi-autonomous platforms
• An eagerness to work across the decision making stack, even on components that may not perfectly align with past experience

#LI-AW1