Machine Learning Engineer ABOUT THE JOB Company Intro: TurbineOne is a fast-moving and high-performance startup with a mission to strengthen situational awareness for all Americans serving at our nation's frontlines - and we are backed by the best DefenseTech venture capitalists. Our Frontline Perception System is an edge-first software platform which allows anyone, even with no technical knowledge, to build and use machine learning models within a comms-contested tactical environment.
Job Title: Machine Learning Engineer Reporting directly to the Chief Technology Officer Geographically flexible for home-office Primary Responsibilities: Solve product challenges from prototype concepts through robust delivery to customers. Iterate with the engineering and product team every step of the way for quick feedback and iteration cycles. Ideate on novel approaches to running state of the art ML systems under compute resource constrained conditions, such as on-device or intermittent cloud connectivity. Creating container-based, production-grade machine learning systems that are robust in field deployments and easy to debug offline. Design T1's Third Party API for Open Source models, allowing other entities to integrate their ML solutions with T1's orchestration engine. Develop infrastructure and processes for growing and curating a custom built solution for ground-truth data sets. Create, maintain and monitor cloud-based systems for continuous quality testing of TurbineOne's ML components over time, as the code and ground truth evolve. Develop unit and integration level testing frameworks for confirming continuous functionality of ML components. Stay abreast of state-of-the-art AI/ML techniques through publications, newsletters, and other means of learning, and communicate such knowledge to coworkers, collaborators, and customers. On a Typical Day You Would: Propose a novel approach to solving specific customer problems, such as improving the ease of labeling or the accuracy of a model on small objects. Propose a way to test the approach given some set of ground truth, and perhaps a process for gathering the ground truth itself. Create a Jupyter notebook to test an approach to a given problem on a standard ground truth dataset and present the findings to the rest of the engineering team, providing a data driven approach to product development. Debug a dip in a ViT model's precision/recall on an object detection dataset as alerted to by the automated quality test suite. Desired Experience and Attributes: High standard of ethics, grit, integrity and moral character. 5+ years of work experience; specifically in moving an idea from a promising experimental result to a production system and/or process. Proficiency in writing documents that lay out experimental results and proposed solutions. Strong software engineering skills with experience in building maintainable complex systems. College degree in Computer Science or Machine Learning. Extensive knowledge of deep learning algorithms and experience in creating and training custom neural network models in TensorFlow, PyTorch, Jax. Experience in optimizing data pipeline and data cleaning/normalization techniques. Startup Culture Expectations: We're a small, fully remote team and everything is our responsibility. Our team thrives on autonomy, trust and solid communication. Everyone on the Team needs to be very comfortable with constant change, moving fast, sharing failures, embracing grit, and building things themselves. Eligibility: Must be eligible to obtain a clearance with the U.S. government