Staff AI/ML Architect, Embodied AI

Intuitive•Sunnyvale, CA

3d•Onsite

About The Position

We are building advanced augmented dexterity capabilities for next-generation robotic platforms. As a Staff AI/ML Architect, you will own the end-to-end architecture of our applied-AI system: a hierarchical, multimodal stack in which a high-level model interprets sensory observations and produces structured intent, and a low-level policy turns that intent into precise, safe, real-time control. You will set the technical direction the rest of the AI/ML team builds against, define the interfaces between perception, reasoning, and control, and make the architecture decisions that let us move safely from offline research to real-time deployment. Working within Intuitive's Future Forward research organization, you will identify, build and finetune AI/ML models and algorithms and define the architecture that enables us to deliver safe and performant embodied AI systems. This role calls for someone who is equally comfortable getting hands-on with models and data and designing systems that scale.

Requirements

MS or PhD in CS, EE, Robotics, or a related field with 10+ years building AI/ML models and algorithms for autonomous-systems software, with demonstrated ownership of an end-to-end AI/ML system architecture taken to production or to a rigorous prototype.
Deep experience with hierarchical, modular AI/ML or robot-learning stacks, including high-level reasoning paired with low-level control policies.
Strong grasp of real-time and safety-critical ML: latency budgets, failure modes, fallback/abort behavior, and interface/contract design across components.
Hands-on expertise with modern deep learning and multimodal, vision-language, LLM-based architectures, including vision foundation models (VFM), vision-action (VA) , and vision-language-action (VLA) models.
Fluency in one of PyTorch, TensorFlow, JAX, and in Python and C++, is required.
Working knowledge of world-model and self-supervised predictive architectures (e.g., JEPA-style models, MAE, DINO) and how learned world models inform perception-and-control design.
Solid foundations in linear algebra, probability, and optimization, enough to reason about and debug model behavior from first principles.
Proven cross-functional technical leadership (able to align research, engineering, and product stakeholders around an architecture).

Nice To Haves

Background in healthcare, medical devices, surgical robotics, or other regulated technical domains.
Foundation-model adaptation and fine-tuning for embodied robotics tasks.
Experience delivering AI/ML in a real-time, safety-critical domain.
Imitation learning, DAgger, and/or reinforcement learning at system scale.
Sim-to-real workflows and experience with robotics simulators (e.g., NVIDIA Isaac Sim).
Regulatory-aware AI/ML development for regulated, safety-critical industries.
Triton kernel and/or CUDA development experiences
Publications or recognized contributions at venues such as CVPR, NeurIPS, CoRL, RSS, or ICRA.
Awareness of data governance in regulated environments (HIPAA, FDA).

Responsibilities

Define and own the end-to-end architecture for a hierarchical perception, reasoning, and control AI system.
Specify the contracts between layers: model outputs, policy interfaces, timing budgets, and safety hooks, and keep them stable as components are swapped and upgraded.
Make the build and modular-vs-monolithic calls, evaluate in-house approaches against the state of the art and set the target architecture.
Establish the path from offline evaluation on recorded data to real-time integration, including the continuous-improvement (human-in-the-loop) data loop.
Partner across research, engineering, data, and product teams, mentor senior scientists/engineers and raise the engineering bar across the effort.
Partner with AI/ML researchers, robotics, data engineers, and other stakeholders to deliver a modular architecture that enables rapid prototyping and learning while working toward a product solution.