Senior Agentic Planning Engineer

As Senior Agentic Planning Engineer, you will own how LightSpeed's robots decide what to do next. Every robot action is one of two primitives, move or actuate. You will build the planner that reasons over a goal, sequences those primitives, chooses classical control or a learned policy for each step, and recovers when a step fails.

Responsibilities:

Primitives and Contracts

• Define primitives and their contract: inputs, preconditions, success checks and failure codes
• Define the interface between the planner and the executor: step request, choice of classical or policy, result and evidence
• Keep the primitive library callable by the planner, not wired by hand for each task

Planning and Reasoning

• Build the task graph from BIM: the elements in front of the robot and the dependencies between them
• Build the planner in two layers: a reasoning layer for open-ended decisions, and solvers for reach, collision and trajectory
• Decide when to use a reasoning model and when not to; keep it off the real-time control path
• Set the rule for choosing classical control or a learned policy for each step, based on world-model confidence and how structured the step is
• Extend the planner's reasoning and sequencing for each new trade
• Measure success by how fast a new trade can be taught

Execution and Recovery

• Build the execution monitor: verify each step, then retry, switch executor, skip and flag, or escalate to a person
• Log every decision with its reason and evidence, so any outcome can be explained
• Keep the planner inside guardrails: a constrained action space, declared parameter ranges, and safety that is never the planner's to decide
• Build the operator view for escalations and remote assistance

Multi-Robot Readiness

• Design the planner so two robots can share one world model and divide work: one holds, one fastens; one supplies, one installs

Foundation and Skill Pipeline

• Co-own the foundation interfaces with the world sensing and robot learning principals, and own the primitive contract
• Run the new stack alongside production on our existing robots, switchable per job, so it is tested on real work from the start
• Prove the system by running whole tasks end to end on real robots, with recovery

Technical Leadership

• Set planning and autonomy architecture across both robot families
• Mentor autonomy and motion planning engineers
• Make the system something engineers can teach new trades with, without you in the loop
• Stay hands-on: write production code and test on real robots every week

Requirements:

• 5+ years building autonomy, task planning or decision-making systems for robots or autonomous vehicles, with systems deployed in production
• Task-level planning on real robots: behavior trees, task and motion planning (TAMP), HTN or similar
• Execution monitoring and fault recovery in a deployed system
• Built interfaces between planning, perception and control that other teams built against
• Hands-on with reasoning models or LLMs in a decision loop, and a clear view of where they help and where they don't
• Strong Python and C++
• Experience with ROS2, Linux, Docker and CI/CD workflows
• A record of setting technical direction across teams and mentoring senior engineers