Computer Vision Engineer

Computer Vision Engineer (Robotics Data & Tooling) About Mecka AI Mecka AI is building the data infrastructure layer for robotics and embodied AI. We work with leading robotics companies and AI labs to collect, label, and validate large-scale real-world visual data used to train perception, manipulation, and control systems. Our work sits directly in the loop between raw sensor data, labeling pipelines, and deployed models. About the Role We’re looking for a Computer Vision Engineer to help us build the internal systems, tools, and models that power our robotics data platform. This role is highly hands-on and product-oriented. You’ll work across: • Computer vision models • Large-scale visual datasets • Internal tools and interfaces for inspecting, debugging, and improving data quality If you enjoy building Gradio / Rerun / custom CV tooling, care deeply about data quality, and want your work to directly impact real robots — this role is for you. What You’ll Do • Build and maintain computer vision models used internally for: • Pre- and post-processing of visual data • Data validation and quality checks • Automated labeling assistance • Design and ship internal CV tools to: • Visualize images, video, and annotations • Inspect failure cases and edge conditions • Explore datasets at scale (filtering, slicing, comparisons) • Create interactive interfaces that help ops, annotators, and researchers: • Understand where data breaks • Identify labeling errors • Decide what data to collect next • Work closely with: • Robotics customers • Data labeling and QA teams • Other CV and ML engineers • Prototype quickly, then harden systems that prove useful into production pipelines What We’re Looking For Must-Have • Strong background in computer vision (images + video; robotics experience a big plus) • An intuitive grasp of linear algebra • Thoroughly grounded in the first principles of traditional CV • Proven track record of delivering high-quality code. • Demonstrated expertise in software development • Consistently produces clean, efficient, and scalable code. • Experience building practical CV systems, not just training models once • Comfortable working with noisy, real-world data • Comfortable shipping internal tools and iterating with users Strong Plus • Experience building CV tooling or apps such as: • Dataset browsers • Annotation / QA tools • Model debugging or visualization dashboards • Gradio-style demos or internal research tools • Familiarity with: • Robotics perception data • Video pipelines (OpenCV, FFmpeg, etc.) • 3D data (point clouds, depth, trajectories) • Exposure to ML infra or data pipelines at scale Tech Stack (indicative) • Python • PyTorch and CV libraries • Video and image processing pipelines • Internal web tooling (React / Typescript or similar) • Cloud-based data infrastructure Exact stack matters less than your ability to build and ship. What Success Looks Like • Internal teams rely on your tools daily • Data quality improves measurably because of systems you built • CV models and labeling pipelines become faster, cheaper, and more reliable • Customers trust Mecka’s data because the underlying tooling is rock solid Who This Role Is Not For • Pure research roles with no production ownership • Frontend-only engineers without CV depth • ML engineers who don’t care about tooling, visualization, or data quality Why This Role at Mecka • Direct impact on how real robots are trained • High ownership over core data and CV systems • Close collaboration with robotics companies and AI labs • Build the tooling layer that most teams wish they had