Robotics Engineer

About Turing

Based in San Francisco, California, Turing is the world’s leading research accelerator for frontier AI labs and a trusted partner for global enterprises looking to deploy advanced AI systems. Turing accelerates frontier research with high-quality data, specialized talent, and training pipelines that advance thinking, reasoning, coding, multimodality, and STEM. For enterprises, Turing builds proprietary intelligence systems that integrate AI into mission-critical workflows, unlock transformative outcomes, and drive lasting competitive advantage.

Recognized by Forbes, The Information, and Fast Company among the world’s top innovators, Turing’s leadership team includes AI technologists from Meta, Google, Microsoft, Apple, Amazon, McKinsey, Bain, Stanford, Caltech, and MIT. Learn more at www.turing.com

This is a remote role, with travel required both within the US & internationally.

Overview

Turing is building the data layer for Physical AI, across four key pillars: (1) Synthetic Data (Sim), (2) Robot Teleoperation, (3) Human Motion Capture, and (4) Data Enrichment.

We are seeking a Robotics Engineer who has first-hand experience training foundation models for robotics. You will not just be an engineer; you will be the architect of our datasets. You know why models succeed or fail in the real world, and you will use that empirical knowledge to design high-value, off-the-shelf datasets that solve those failures for our customers.

You will act as the bridge between our data collection teams and our customers' model performance, eventually building out a "Benchmarking & Optimization" business unit that evaluates, diagnoses, and improves the performance of our customers’ robotics foundational models.

Why This Role?

• Help Launch Business Unit: a 0-to-1 opportunity to help build a new business unit within Turing.
• Define Industry Standards: Architect off-the-shelf datasets that will fuel the robotics ecosystem, setting the benchmark for Physical AI data quality.
• Own the Evaluation Loop: Lead the critical feedback loop that validates data quality, proving empirically how specific attributes drive model performance.

Key Responsibilities

1. Data Strategy & Productization (Short-Term Focus)

• Design "Off-the-Shelf" Solutions: Leverage your experience to define specifications for high-value dataset products (e.g., "The definitive bi-manual manipulation dataset for household objects"). You will dictate the requirements to our Sim, Teleop, and Motion Capture teams to ensure they capture the right data.
• Commercial Engagement: Partner with the Sales team as a key technical authority. You will translate customer pain points (e.g., "Our model fails on transparent objects") into well defined data solutions.
• Quality Definition: Move beyond basic QA (resolution, frame rate) to semantic QA. Define what "high-quality" data looks like for model learning.

2. Model Evaluation & Gap Analysis (The Technical Core)

• Build the "Eval" Loop: Establish workflows to train and fine-tune open-source Robotics Foundation Models (e.g., OpenVLA, Octo, RT-X) on our data. You will empirically prove that Turing’s data improves model performance.
• Agentic Gap Analysis: Develop automated workflows to analyze datasets for coverage gaps. Use this analysis to provide high-value insights to customers (e.g., "Your model is failing because your training distribution lacks diverse lighting conditions—here is the synthetic data to fix it").

3. The Future: "End-of-Cycle" Optimization (Long-Term Focus)

• Benchmark-as-a-Service: Lead the development of virtual and physical benchmarking environments to test customer policies.
• The Optimization Loop: Build a full-cycle service where Turing: (1) Benchmarks a client's model, (2) Identifies performance gaps, (3) Translates gaps into data requirements, and (4) Delivers the data to close the loop.
• Establish a New Business Unit: Scale these capabilities into a dedicated "Benchmarking & Optimization" business unit. You will help build this new pillar, recruiting the team, defining the product solution, and driving the monetization strategy.

Required Qualifications

• 8+ years of total experience as a seasoned technologist with cross-functional operational experience.
• 5+ years in Robotics Learning with deep hands-on experience training and deploying machine learning models for robotics (Imitation Learning, RL, or VLA architectures).
• 3+ Years of cross-functional & commercial experience in which you have "bridged the gap" between research and product sales (e.g., Applied Scientist, Technical Lead, or Solutions Architect).
• Expert proficiency in Python. You must be capable of writing production-grade ML code, custom data loaders, and evaluation pipelines from scratch.
• Deep proficiency in PyTorch or JAX. Experience implementing and debugging complex deep learning architectures.
• Direct hands-on experience training, fine-tuning, or evaluating advanced robotics policies. Specifically, familiarity with Diffusion Policies or VLA architectures.
• Proficiency with ROS (Robot Operating System). You must understand message passing, node architecture, and how to interface with hardware drivers.
• Experience manipulating standard robotics data formats. You understand the structure of "trajectory" data (state-action pairs).
• Strong grasp of 3D spatial mathematics (quaternions, coordinate frames, transformations) essential for manipulating robot kinematics and camera data.
• Entrepreneurial mindset required: ability to own end-to-end solutions, including technical execution, customer engagement, cost/ROI tradeoff analysis, and value communication to technical and non-technical stakeholders.

Preferred Qualifications

• You know exactly what makes a dataset valuable when building advanced robotics foundational models.
• You are comfortable explaining technical constraints to Sales teams, Product Managers, or external clients.
• Experience with NVIDIA Isaac Sim, Unity, MuJoCo, or Genesis. Ideally, you have built custom benchmark environments or generated synthetic datasets.
• Experience setting up distributed training workflows on cloud clusters and using experiment tracking tools like Weights & Biases (W&B).
• Experience using Vision-Language Models (VLMs) and other tools to automatically label or annotate large datasets.
• Ability to read and debug C++ code is a plus (useful for troubleshooting hardware drivers or legacy ROS nodes), but production coding in C++ is not required.
• Experience deploying models to edge compute devices (e.g., NVIDIA Jetson) or directly onto robot controllers.