Head of Autonomy

Teleo, a Havoc company, is a robotics company that transforms construction heavy equipment, including loaders, dozers, excavators, and trucks, into autonomous robots for commercial and defense applications. Our technology enables a single operator to supervise and control multiple machines simultaneously, delivering significant productivity gains while improving operator safety and comfort.

Teleo was founded by a team of experienced technology leaders who previously led the development of Lyft's Self-Driving Car program and Google Street View. Teleo recently announced its merger with Havoc AI, a fast-growing defense technology company developing coordinated fleets of autonomous maritime vessels.

This is a unique opportunity to join a team building technology with real-world impact. You will work on cutting-edge 100,000-pound autonomous robots and engineer complex systems at the intersection of hardware, software, robotics, and AI.

The Role

The Head of Autonomy owns Teleo's autonomy stack end to end and leads the team that builds it. You will lead the engineering team across perception, simulation, controls, path planning, fleet orchestration, learning-based methods (RL, imitation learning, and VLA-style models), and MLOps.

This is a senior player-coach role. You will set the technical direction, make the architecture calls, and stay close enough to the code and the field data to review designs in depth. You will also hire, grow, and run the team, and be accountable for autonomy performance on customer deployments across several machine types.

The core technical problem is taking a stack that works today in supervised autonomy on real job sites and scaling it: more machine types, more sites, more material-manipulation tasks, and fewer operator interventions per hour, without compromising safety.

Core Responsibilities

Technical leadership

• Own the autonomy architecture from sensors to actuation: perception, localization, world modeling, planning, control, and the interfaces between them
• Set the roadmap for moving from hand-engineered components to learned ones (imitation learning, RL, VLA-style policies), and decide where classical methods should stay
• Drive a composable, skill-based approach to material manipulation (loading, pushing, scooping, dumping, digging) that transfers across machine types
• Build the simulation and data engine: operator data capture, sim-to-real validation, closed-loop evaluation, and regression testing tied to field metrics
• Scale autonomy to multi-machine fleets working alongside remote operators, including task allocation and coordination on shared sites
• Shorten the time to bring autonomy up on a new vehicle model, working closely with hardware and vehicle integration