Job description
Overview
Innomium is moving into robotics, and we’re looking for an exceptional Embodied AI Engineer to join us.
We are building toward intelligent systems that can perceive, reason, learn, and act in the physical world. We’re looking for engineers and researchers with deep hands-on experience in Embodied AI, robotics, multimodal intelligence, simulation, data generation, benchmarking, and learning-based control.
This is not a traditional software engineering role. We are especially interested in people who have worked on systems where AI interacts directly with physical or simulated environments.
What You’ll Work On
You may work across areas including:
Embodied AI and robotic intelligence
Vision-Language-Action (VLA) models
Robot learning and manipulation
Reinforcement learning and imitation learning
Learning from demonstrations and human feedback
Multimodal perception and reasoning
Vision-language models for robotics
Navigation, planning, and control
World models and spatial intelligence
Sim-to-real and real-world robot deployment
Robot foundation models
Training and evaluating embodied agents
Integration of AI models with robotic hardware and sensors
Creating high-quality simulation environments for training and evaluation
Designing new benchmarks and evaluation frameworks for embodied intelligence
Generating synthetic, simulated, and real-world training data
Building scalable data-generation pipelines for robotic learning
Creating tasks, environments, scenarios, and evaluation suites to measure model capabilities
Developing simulation infrastructure for large-scale robot training, testing, and experimentation
We value people who can move between research, experimentation, simulation, data, engineering, and real-world deployment.
What We’re Looking For
Strong candidates may have experience with some combination of:
Embodied AI research or development
Robotics and autonomous systems
Manipulation, locomotion, or navigation
Vision-Language-Action models
Reinforcement learning, imitation learning, or behavior cloning
Computer vision and multimodal models
Transformer-based architectures and foundation models
PyTorch or similar deep-learning frameworks
ROS / ROS2 and robotics software stacks
Robot simulation environments such as Isaac Sim, MuJoCo, Gazebo, Habitat, or similar platforms
Building custom simulation environments and robotic tasks
Generating synthetic datasets or large-scale robotic training data
Designing benchmarks for manipulation, navigation, planning, reasoning, or general embodied intelligence
Creating automated evaluation and testing pipelines
Domain randomization, procedural environment generation, and sim-to-real transfer
Training models using large-scale robotic or multimodal datasets
Deploying learned policies on physical robots
Sensors, cameras, depth perception, proprioception, and robot control
Academic credentials are welcome, but demonstrated ability and real experience matter more to us than titles.
How to Apply
Please send us a detailed description of your experience and capabilities in Embodied AI.
We would especially like to understand:
What Embodied AI or robotics systems you have personally built or worked on
Your exact contribution to each project
The models, architectures, algorithms, and training approaches you used
What robotic platforms, sensors, or simulation environments you have worked with
Whether you have created your own simulation environments, tasks, or scenarios
Whether you have designed benchmarks or evaluation frameworks for embodied agents
How you have generated, collected, cleaned, labeled, or scaled training data for robotics or embodied AI
Whether you have built synthetic-data or simulation-based data-generation pipelines
Your experience with VLA models, robot foundation models, reinforcement learning, imitation learning, or related approaches
Whether you have deployed AI models on real robots
The most difficult technical problems you encountered and how you solved them
Research papers, publications, GitHub repositories, demos, videos, datasets, benchmarks, or projects that demonstrate your work
What you believe you can contribute to building advanced embodied intelligence at Innomium
Please be specific and technical.
Rather than simply listing technologies, we want to understand what you have actually built, what you personally accomplished, and the depth of your ability in Embodied AI.
If Embodied AI and the future of intelligent robotics are what you want to spend your time building, we’d like to hear from you.
How to Apply
Please submit:
- Your résumé or professional profile.
- Links to relevant GitHub repositories, products, models, evaluations, technical writing, design work, campaigns, or other inspectable evidence.
- A brief explanation of a system, product, model, or program you meaningfully owned—your role, the decisions you made, and the outcome.
- Your location, availability, and preferred working arrangement.
We are more interested in clear evidence of ownership, judgment, and craft than in an extensive list of technologies. Generic cover letters are not required. Compensation, eligibility, and working-time overlap are confirmed early in the process.