Principal Robotics Machine Learning Engineer
About this role
Apptronik needs a Principal Robotics Machine Learning Engineer to solve the critical challenge of enabling humanoid robots to interact seamlessly with the physical world through advanced manipulation capabilities. The engineering team builds software and AI systems for the flagship humanoid robot, Apollo, with an initial focus on manufacturing and logistics environments. This person will lead the end-to-end lifecycle of learned manipulation, bridging the gap between research concepts and robust production software running on physical hardware.
Working closely with data infrastructure and hardware teams, the engineer will design data engines, train machine learning models, and deploy solutions directly onto robotic platforms. Success in this role requires deep technical fluency in both classical computer vision and modern deep learning techniques, combined with a pragmatic startup mindset. Candidates who thrive here possess high ownership, value fast feedback loops, and demonstrate a strong commitment to mentoring others while scaling robotics technology to market.
What you will do
- Implement and deploy advanced machine learning models and algorithms to execute open-world object manipulation tasks on physical hardware.
- Manage the end-to-end data engine for manipulation learning, encompassing data collection strategies, curation, annotation, and the balance of synthetic and real data.
- Drive the complete development cycle from simulation prototyping to robust model transfer and fine-tuning directly on the robot.
- Diagnose and isolate performance bottlenecks by distinguishing between root causes in classical vision and sub-optimal model or data configurations.
- Mentor engineers, conduct code reviews, and establish high standards for architectural rigor across the organization.
What you need
- Bring at least five years of professional experience shipping machine learning models on robotic systems within production environments.
- Demonstrate expertise in AI model development, including proficiency with convolutional neural networks and transformer architectures.
- Show strong software engineering capabilities using Python, C++, and PyTorch to build real-time robotic software stacks.
- Possess hands-on experience developing AI data pipelines for collection, annotation, and synthetic data generation.
- Apply deep familiarity with computer vision concepts such as 6D pose estimation, camera calibration, point cloud processing, and visual-servoing.
- Accumulate seven years of relevant experience, or five years with a PhD, specifically focused on robotic manipulation or complex motion control.
- Hold a Bachelor's, Master's, or PhD degree in Robotics, Computer Science, Electrical Engineering, or a related discipline.
Nice to have
- Familiarity with distributed training and MLOps principles
- Experience with VR, haptic interfaces, and retargeting algorithms for human-in-the-loop teleoperation control
- Proficiency with physics engines such as IsaacSim, MuJoCo, or Drake for policy training and validation
- Background in logistics automation using manipulators for dynamic real-world material handling tasks
Skills and tech stack
Questions candidates ask
What level of experience is required for this role?
Candidates must have 7+ years of relevant experience, or 5+ years with a PhD, specifically focused on robotic manipulation or complex motion control, along with 5+ years shipping machine learning models on robotic systems in production.
What technologies and frameworks will I use?
The tech stack includes Python, C++, PyTorch, CNNs, and transformers, with additional exposure to simulation engines like IsaacSim, MuJoCo, or Drake.
What does the development lifecycle involve for this position?
The role covers the end-to-end lifecycle of learned manipulation, from data engine design through training to on-board robot deployment and performance bottleneck diagnosis.