Software Engineer, Data Infra

Job description

Dyna Robotics makes general-purpose robots powered by a proprietary embodied AI foundation model that generalizes and self-improves across varied environments with commercial-grade performance. Dyna's robots have been deployed at customers across multiple industries. Its frontier model has the top generalization and performance in the industry. Dyna Robotics has raised over $140M, backed by top investors including CRV, First Round Capital, Robostrategy, Salesforce Ventures, NVentures, Amazon, Samsung Next, and LG Technology Ventures. Our team brings together engineers and researchers from Google, Meta, Apple, Amazon, Cruise, Aurora, NVIDIA, along with academic roots at Stanford, Berkeley, MIT, UPenn, and beyond. We're positioned to redefine the landscape of robotic automation. THE ROLE As a Software Engineer, Data Infra you are the architect of the "Laboratory" where Dyna’s robotic intelligence is refined. You won't just move data; you will build the interactive systems that bridge the gap between raw multimodal sensor streams and production-ready ML models. This is a high-impact, hands-on role where you will design the infrastructure to visualize model performance, automate data labeling, and manage the "human-in-the-loop" workflows - such as teleoperation and skill capture - that are critical to our scaling. If you enjoy solving complex geometric problems and building the low-friction tools that empower ML researchers, you’ll thrive here. WHAT YOU’LL DO - Interactive Data Systems: Architect the engines and interfaces that unify raw robot logs, video, and 3D sensor data. You will enable seamless "human-in-the-loop" workflows, from episode annotation to analyzing manual interventions. - Signal Extraction & Geometry: Design and implement algorithms to extract structured signals (trajectories, events, 3D poses) from raw captures. Strong mathematical intuition is required to turn pixels and point clouds into ground-truth insights. - Evaluation & Benchmarking: Build high-performance tools to compare model-driven motion against human-captured data, helping the team quantify model progress across diverse tasks. - Scalable ML Pipelines: Build and operate distributed data pipelines (using Python, GCP/AWS, and Kubernetes) for the ingestion, transformation, and validation of terabytes of multimodal data. - Observability & Debugging: Develop visualization tools that make complex model behaviors and sensor data easy to interpret, reducing the time from "data collected" to "model trained." - Startup Fluidity: Collaborate across ML, Robotics, and Product teams. As an early member of the data team, you will help define the roadmap where no blueprint yet exists. WHAT YOU’LL BRING - The Experience: 5+ years of professional software experience, ideally with a focus on data-intensive or "human-in-the-loop" platforms. - Technical Stack: Proficiency in Python (NumPy, Pandas) and a solid understanding of modern backend architectures. Experience with React/TypeScript is a major plus for building internal observability tools. - Mathematical Core: Strong skills in algorithms and geometric calculations (e.g., coordinate transformations, 3D spatial reasoning). - Data Mastery: Hands-on experience with relational and NoSQL databases (PostgreSQL, Redis) and cloud-native infrastructure (GCP/AWS). - Problem-Solving: The ability to debug real-world data issues - from sensor drift to pipeline bottlenecks - independently. BONUS POINTS FOR: - Experience with multimodal data (video, LiDAR, time-series) in robotics or autonomous systems. - Familiarity with Airflow, Kubeflow, or similar distributed batch processing systems. - Experience with experiment tracking frameworks like Weights & Biases or MLFlow. - Experience as an early hire in a fast-paced startup environment. At Dyna Robotics, we build technology for the real world, which requires a team as diverse as the environments our robots inhabit. We are an equal opportunity employer committed to technical rigor and mutual respect. Don't let a checklist stop you. Data shows that underrepresented groups often only apply if they meet 100% of the criteria. We value problem-solving and grit over keyword matching. If you're passionate about closing the loop between deployed robots and better models, we want to hear from you, even if you don't check every box.