Machine Learning / Robotics Engineer — Simulation

Marvik; United States; Remote; USD 120,000 - 180,000; Full time

Job summary

Marvik is seeking a Machine Learning or Robotics Engineer with strong simulation background to verify AI in a safe, simulated environment before live deployment. You will generate synthetic data, build end-to-end test pipelines, and connect AI models to sensors and hardware for real-world validation.

You will collaborate with engineers, researchers, and clients to translate open-ended requirements into practical, scalable solutions, while traveling internationally for on-site support when needed.

Qualifications

• Professional experience in machine learning engineering or robotics.
• Hands-on time with simulation platforms like Isaac Sim/Isaac Lab, MuJoCo, Gazebo, or PyBullet.

Responsibilities

• Build simulated environments to test AI systems before real deployment.
• Generate synthetic data and create repeatable pipelines from data collection to training and evaluation.
• Connect models to sensors, software, and hardware when required and design experiments to validate in simulation and reality.
• Take experiments from prototype to production integration and develop reusable tooling.
• Possibly travel internationally for on-site client integration.

Skills

Python

C++

ROS/ROS2

GPU optimization

Docker

Cloud infrastructure

Tools

MuJoCo

Gazebo

PyBullet

NVIDIA Jetson

Job description

Описание:

Most AI gets tested on a screen. Yours will get tested in a simulator first, and then in the real world. At Marvik, we’re looking for a Machine Learning or Robotics Engineer who has spent real time in simulation and wants to see that work leave it. You’ll join a project with a top-tier global client, building AI that interacts with the physical world. What you get in return is real ownership and the chance to watch your work run in real-world conditions.

Задачи:

Your job is to make sure AI systems know what they’re doing before they meet reality. You’ll build the simulated environments where algorithms get tested, synthetic data gets generated, and failures show up early, plus the pipelines that make every result repeatable, from data collection to training and evaluation. You’ll connect models to sensors, software, and hardware when the project needs it, and design the experiments that tell us how much of what worked in simulation survives outside it (and why, when it doesn’t). You’ll take things from the first experiment all the way to production integration, working side by side with engineers, researchers, and the client to turn open-ended requirements into practical solutions. What proves useful becomes reusable tooling that makes the next project faster. And sometimes it means traveling internationally to support on-site integration.

Требования:

It’s a big plus if you’ve worked with computer vision, 3D perception, or multimodal learning; synthetic data, domain randomization, or sim-to-real evaluation; reinforcement learning, imitation learning, motion planning, or control; ROS / ROS 2 with physical sensors or robots; or edge deployment on devices like NVIDIA Jetson. C++, GPU optimization, Docker, cloud infrastructure, or experience turning prototypes into reliable production systems also help.

Условия:

We’re fully remote, and we expect you to own what you build. That means navigating open-ended problems, explaining trade-offs clearly, and taking responsibility for both implementation and validation. Nobody hands you a ticket and disappears: you’ll be in the conversation with the people who define the problem and the people who depend on the solution. You’ll fit if… You have professional experience in machine learning engineering, robotics, or a closely related field, and hands‑on time with a simulation platform such as NVIDIA Isaac Sim, Isaac Lab, MuJoCo, Gazebo, PyBullet, or something comparable. You write strong Python and care about solid software engineering. You know how to evaluate ML models or robotics algorithms and which metrics actually matter. You’re comfortable with the physical side of the problem (sensor modeling, coordinate transformations, kinematics, dynamics, or physics‑based simulation), and Linux and Git are part of your daily routine. You don’t need to master every corner of the stack. We welcome ML engineers who have used simulation for model development, and robotics engineers who have integrated machine learning. Depth in one area and the drive to work across disciplines matter more than covering everything.