Job Details

Company Overview:

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 was founded by repeat founders Lindon Gao and York Yang, who sold Caper AI for $350 million, and former DeepMind research scientist Jason Ma. The company has raised over $140M, backed by top investors, including CRV and First Round. We're positioned to redefine the landscape of robotic automation. Join us to shape the next frontier of AI‑driven robotics!

Learn more at dyna.co

Position Overview

We are seeking a Research Engineer / Scientist to join our team and help us build and deploy cutting‑edge AI models that work reliably in the real world. This role is a hybrid of research and engineering, requiring you to be both a skilled model trainer and a meticulous debugger. You will be responsible for the full lifecycle of production research pipelines—from training advanced models to analyzing their behavior and driving continuous improvements.

Your work will be crucial for both our short‑term deliverables and long‑term competitive advantage. You will translate ambiguous, real‑world challenges into clear, solvable research and engineering problems, and your research will focus on developing new AI models and algorithms for dexterous robotic manipulation. You will collaborate closely with cross‑functional teams to rapidly iterate on novel algorithms and large‑scale model training to deliver general and adaptive robotic manipulation models.

Key Responsibilities

• Research, develop, and fine‑tune large deep learning models, including vision‑language models (VLMs) and other AI models for robotic manipulation tasks.
• Design, implement, and evaluate new robot learning algorithms, such as reinforcement learning and imitation learning, on both physical robots and simulation platforms.
• Analyze training results, debug model failures, identify performance bottlenecks, and resolve subtle implementation issues to improve overall system performance.
• Run structured experiments and ablations to isolate root causes and inform iterative improvements.
• Collaborate with data and robotics teams to refine data collection, labeling, and filtering strategies.
• Build tools and dashboards for tracking metrics, visualizing model performance, and surfacing anomalies.
• Translate practical product needs or robot behaviors into well‑scoped research or engineering problems.
• Produce and deliver dependable, high‑quality robotic software solutions.

Qualifications

• Minimum of 5 years of experience in applied machine learning, deep learning, or AI and robotics research.
• Master's degree or above in Computer Science, Robotics, or a related discipline, or comparable hands‑on experience.
• Comprehensive understanding of machine learning and deep learning fundamentals, especially with Transformer architectures, representation learning, and robot learning methodologies (e.g., reinforcement learning, imitation learning).
• Practical experience in debugging, optimizing, and improving deep learning models in production or research settings.
• Proficiency in Python and at least one major deep learning framework like PyTorch, TensorFlow, or JAX.
• Familiarity with robot simulation platforms such as Mujoco, NVIDIA Isaac Stack, or Drake, and experience deploying AI models on physical robots.
• Comfortable with data analysis tools and workflows (e.g., NumPy, Pandas).
• Demonstrated ability to turn ambiguous real‑world issues into actionable research/engineering projectsli>
• Superior analytical and problem‑solving abilities with a methodical, detail‑oriented approach to experimentation.
• Strong enthusiasm for robotics and developing robotic products.
• Understands how to write scalable, modular, and maintainable research code in collaborative environments.

Preferred Qualifications

• PhD in Computer Science, Electrical Engineering, Robotics, or a related discipline.
• Experience with robotics systems (ROS) or production inference/debugging infrastructure.
• Familiarity with multi‑modal foundation models.
• Demonstrated research excellence, evidenced by publications in prestigious conferences such as NeurIPS, ICML, ICLR, CVPR, RSS, CoRL, or ICRA.
• Exceptional creativity, productivity, and independent research skills.