Origin is building a general-purpose construction robot designed to automate interior finishing work—drywall, painting, and related trades—on live job sites. The tech stack reveals a company deeply invested in computer vision and embodied AI: PyTorch, JAX, CUDA, and a suite of vision-language models (CLIP, BLIP-2, LLaVA) paired with robotics fundamentals (ROS 2, MoveIt 2, Isaac Sim). Active projects span deep learning for semantic understanding, photorealistic simulation, and auto-annotation pipelines—suggesting the core challenge is training models on real construction data at scale. The hiring mix is engineering-heavy (18 of 27 roles), with balanced senior and lead representation, indicating both execution velocity and mentorship depth needed for hardware-software integration.
Tech Stack 116 technologies
What Origin Is Building
◆Challenges
- Safety risks
- Labour shortages
- Rising construction costs
- Reducing rising costs
- Production readiness
- Lead-time reduction
- Lack of high-quality labeled data
- Reducing construction labor shortages
- Scaling mechanical supply chain
- Mechanical bom cost optimization
▲Active Projects
- Operator training program
- Pcb design for intermediate-complexity boards
- Deep learning model training for semantic understanding
- Photorealistic wall-surface simulation via diffusion models
- Vision-language models linking work orders, cad plans, and sensor data
- Auto-annotation pipelines scaling to millions of frames and point-clouds
- Robot manipulation & control algorithm development
- Motion planning systems
- Navigation system development
- Dataset creation for ai agents
