Perception Engineer

At Terra Firma, we’re redefining how the world moves earth for construction. Founded by former Space

X engineers and backed by Bain Capital Ventures, Terra Firma is automating construction to make it faster, cheaper, and safer.

At Space

X, we were part of the teams that built the largest rocket in human history, launched the largest constellation of satellites ever made, and delivered connectivity to millions of people worldwide. What we learned was simple: infrastructure is everything.

At Starbase, the biggest barrier to launching missions to Mars was not building the rocket. It was how slow, expensive, and complex it was to build the infrastructure around it. We lacked buildings, roads, water, power, communications, and waste management, and constructing them took far more time and resources than it should have.

From a first‑principles perspective, that doesn’t make sense. It shouldn’t be this hard or costly to create the foundations for progress. That’s when we realized something important: the same engineering mindset that enabled us to tackle space exploration and mass manufacturing could be applied to infrastructure. By bringing the principles of simplification, iteration, acceleration, and automation to construction, we can unlock a new era in how the world builds.

Accelerating infrastructure unlocks progress across every domain, creating a self‑sustaining engine for tackling humanity’s toughest challenges. The innovations we develop on Earth will lay the groundwork for the technologies that one day make settlement on Mars possible.

You will design and implement the perception systems that give our automated heavy machinery the awareness needed to operate effectively in complex, unstructured environments. This role combines high‑level architecture decisions with hands‑on algorithm development, hardware integration, and field testing. You will work across the perception stack, from integrating and calibrating advanced sensors to building real‑time processing pipelines for images, point clouds, and other high‑bandwidth data streams.

Your work will power critical functions such as collision avoidance, motion planning, localization, and operator visualization.

• Integrate multi‑spectral and TOF sensors into robotic systems with real‑time processing constraints
• Work with motion and trajectory planning teams to design and synthesize sensor systems to extract required signals and inputs
• Implement algorithms to efficiently detect objects of interest from incoming data streams for the purposes of collision avoidance, motion planning, and trajectory planning
• Stitch together and manipulate camera feeds and data from other sensors to effectively visualize the state of physical systems to human operators
• Develop real‑time pipelines for processing, sharing, and fusing data from multiple sensor sources
• Train Machine Learning (ML) models on existing and custom datasets and support collection and labeling of data for additional training and validation
• Develop simulation and visualization tools to aid in rapid prototyping of robotic software and control algorithms
• Own the integration, calibration, and testing of sensors, both for proof‑of‑concept prototypes and production systems
• Lead interdisciplinary teams, mentor new employees, and actively participate in the hiring process

• B.S. in Robotics, Mechanical Engineering, Physics, Computer Science, Electrical Engineering, or a closely related field
• 2+ years of professional or project experience with robotic software development, involving implementation of real‑time algorithms for perception and data processing pipelines
• Demonstrated experience owning large‑scale deliverables, leading teams, and driving product roadmaps

• Experience working with safety‑critical or real‑time systems
• Experience working with embedded applications for robotics, autonomous vehicles, aerospace, automotive, or construction
• Proficiency in C++ and Python, including development for Linux‑based systems and bare‑metal MCUs
• Experience training, testing, optimizing, and deploying online and offline machine learning…