Senior Perception Engineer (ML)

At Scythe, we're bringing an industry-leading autonomous battery-powered mower to market that runs on a powerful single-board ARM and Linux computer with a custom software stack running everything from sensors stream capture, to autonomy, to logging, to cloud communications and user interactions. As a Perception Engineer focused on ML development, you will build the future of off-road, real-time robot perception. Your primary focus will be the development of novel obstacle detection and scene recognition algorithms tailored especially to the task of safely mowing semi-structured environments, by developing, testing, and deploying perception algorithms that can run in real-time using embedded GPUs.

You’ll work with teammates from across the company to build an understanding of perception needs, brainstorm solutions, and bring those solutions to life. We expect you to be broadly experienced to be able to “see around corners”, to thrive in a fast-paced, self-managed engineering environment, and to be a collaborative, low-ego teammate who helps Scythe realize its big mission.

What you’ll do at Scythe  

• Help maintain, test, and improve existing perception systems
• Develop, test, and deploy scalable and accurate data filtering, surfacing, and selection pipelines to assist the data labeling and QA processes
• Develop, train, test, optimize, and deploy novel real-time machine learning models for a variety of perception tasks, such as semantic and instance segmentation, depth estimation, pose estimation, calibration estimation, keypoint detection and feature extraction, optical flow, etc
• Develop, test, and deploy novel perception algorithms tailored specifically for the purpose of mowing autonomously, such as 3d mapping, drivable surface estimation, obstacle detection, etc
• Write high-quality, performant, and safe code that will run on the robot
• Collaborate with engineers from other specialties (calibration, localization, motion planning, controls) to make sure work is in sync with company-wide goals and current needs

What you know well

• Mathematical concepts relevant to the tasks of training, testing, and optimizing machine learning models, such as deep knowledge of geometry and linear algebra, matrix decomposition, differential calculus, probability theory, and statistics
• Foundations of computer vision such as camera models, multi-camera calibration, image processing, HDR, stereo matching, etc
• Experience using and expanding state-of-the-art deep learning models for obstacle detection, semantic and instance segmentation
• Experience developing, optimizing, and deploying neural networks for vision tasks such as obstacle detection, semantic segmentation, and instance segmentation
• Proficiency in a compiled programming language such as C++, Rust, or Go, and their development in large codebases. We use C++ and CUDA for perception interaction with the GPUs, but Rust for most of our codebase, so you must be open to learning it and adopting it for everyday use
• Implemented, deployed, and integrated different perception algorithms to turn camera pixels into obstacles in real-world robots

What you’ve maybe done

• Developed real-time machine learning-based computer vision models
• Developed scalable data pipelines using active learning to select training data efficiently from a large unlabeled pool of data
• Worked with self-supervised training of vision models for representation learning
• Leveraged simulated data to improve the quality of supervised vision models
• Wrote CUDA code to implement real-time perception algorithms
• Worked with dense stereo matching and 3D reconstruction
• Worked with ROS and its communication scheme

Why Scythe?

• Scythe is an early-stage but well-capitalized startup. Have a huge impact alongside an awesome team of experts shipping something the world has never before seen
• Competitive salary and equity compensation
• Fully-sponsored medical, vision, and dental insurance, including 75% funded dependent coverage
• 401(k) retirement plan (non-matching today)
• Headquarters near beautiful Boulder, CO. Enjoy the bounties of nature and open spaces close to home with mountain biking, hiking, skiing and more.
• Satellite offices in Austin, TX, and Vero Beach, FL 
• Flexible paid time-off and remote work to let you do your best work where and when you want
• Highly collaborative learning culture where personal freedom, growth, and responsibility are valued
• An opportunity to have an incredibly positive impact on the world

Closing

Scythe is a total compensation company, which provides employees with a comprehensive salary, equity, and benefit package. However, only the minimum salary amounts are listed here. Scythe carefully considers a wide range of compensation factors, including education, years of experience, competencies and other relevant business considerations. These considerations can cause your compensation to vary along with your compensation mode preferences. The Senior Perception Engineer (ML) position has an expected minimum annual cash salary of $140,000. The actual pay may be higher depending on your skills, qualifications, and experience. Equity and benefits packages are NOT included in this estimate. Please note that this information is provided for those hired in Colorado only, and this role is open to candidates outside of Colorado as well.

Scythe is an Equal Opportunity employer committed to diversity in the workplace. All qualified applicants will receive consideration for employment without regard to race, color, religion, age, national origin, sex, sexual orientation, gender identity, disability, protected veteran status or any other factor protected by applicable local, state or federal laws.