Lead Scientist

About Origin

Origin (previously 10xConstruction) is building general-purpose autonomous robots for US construction to tackle rising costs, safety risks, and labour shortages. Our modular, multi-trade platform combines purpose-built hardware with real-time site intelligence to navigate complex environments and execute tasks with precision. Trained in high-fidelity simulation and already deployed on live sites, our robots deliver 5x faster execution, 250%+ margin expansion, and significant cost savings. Join India's most talent-dense robotics team consisting of individuals from IITs, Stanford, UCLA, etc.

About the Role

You will own the quality and performance of every tool the robot operates: spray guns, sanders, and future finishing tools used in indoor construction. This means running rigorous experiments to find the right operating parameters, characterising defects, building mitigation strategies, and producing the structured datasets the AI team needs to close the loop on autonomous tool use. You will lead a cross-disciplinary team spanning manipulation, perception, AI, and mechanical engineering, and you will build the physical testing infrastructure (sample panels, mock environments, instrumented rigs) that makes all of this possible.

Key Responsibilities

• Design and execute structured, multi-parameter experiments (Design of Experiments methodology) to identify optimal tool parameters (pressures, speeds, distances, feed rates, material combinations) for each finishing application. Isolate variables correctly, document methodology and controls, and report results with statistical significance.
• Characterise defect modes (runs, sags, orange peel, uneven sanding, missed coverage) across applications; develop and validate mitigation strategies with measurable pass/fail criteria. Understand when results show real causal relationships versus spurious correlations
• Own the vision-based data collection pipeline end-to-end: select cameras, lenses, and lighting setups; define image capture sequences and viewpoints; collect, process, and quality-check image data; and deliver labelled, versioned datasets structured for direct consumption by the AI team.
• Lead the cross-disciplinary applications team (manipulation, perception, AI, mechanical) to deliver validated tool-use capabilities for spraying, sanding, and subsequent indoor construction applications.
• Program and operate robot arms (via pendant interface) to execute tool-use experiments; collaborate with the manipulation team on trajectory design and controller tuning for each application.
• Specify, procure, and commission the testing infrastructure: sample substrates, mock wall assemblies, spray booths, dust extraction, camera/lighting rigs, and instrumented measurement setups.
• Own the end-to-end operations of the applications lab: scheduling, safety protocols, consumables inventory, equipment calibration, and the construction/destruction cycles required to maintain a fresh supply of drywall test environments.

Required Qualifications and Skills

Experimental Research

• Formal training or deep working experience in experimental design: Design of Experiments (DOE), controlled multi-parameter studies, statistical analysis, and data interpretation. You understand that you only change the variable you are studying, you know when a result is statistically significant, and you know that correlation is not causation.
• Ability to fit curves and build empirical models, but more importantly, to critically evaluate what those models do and do not explain. Comfortable with Python for data analysis, visualisation, and automation of experimental workflows.

Vision-Based Data Collection

• Hands-on experience designing and executing vision-based data capture: camera and lens selection, lighting design, image sequence planning, and capture execution. Able to reason about what makes an image dataset complete, consistent, and useful.
• Working knowledge of image processing fundamentals: colour spaces, exposure, distortion, basic filtering and annotation. You do not need to train models, but you need to produce the data that makes training possible

Robotics

• Understanding of modern robotics systems: robot arms, trajectory planning, controllers, and how mechanical design and electronic systems integrate into a working platform. Working proficiency with industrial manipulators through their pendant/teach interface (jogging, waypoint programming, routine execution).
• Hands-on experience in at least one robotics-adjacent vertical (manufacturing, automation, research) or demonstrated exposure to the interplay between mechanical, electrical, and software subsystems on a robotic platform.

Preferred Experiences

• Understanding of AI/ML models, especially vision-based architectures (object detection, segmentation, defect classification), with enough depth to make informed decisions about what data the models need and how collection choices affect model performance.
• Product thinking around construction tools and finishes: familiarity with drywall, plastering, painting trades, commercial quality standards, and the practical realities of what "good" looks like on a job site
• A relevant PhD (materials science, robotics, applied physics, computer vision, or a related experimental discipline) that reflects deep comfort with scientific methodology and independent research.
• Operations experience driving construction and destruction cycles of a physical test environment, including the logistics of standing up drywall assemblies, running experiments, tearing down, and repeating, alongside the data collection and labelling operations that run in parallel