Robotics Researcher - Reinforcement Learning

Apptronik is building robots for the real world to improve human quality of life and to help solve the ever-increasing labor shortage problem. Our team has been building some of the most advanced robots on the planet for years, dating back to the DARPA Robotics Challenge. We apply our expertise across the full robotics stack to some of the most important and impactful problems our society faces, and expect our products and technology to change the world for the better.  We value passion, creativity, and collaboration to help us overcome existing technological barriers in the industry to create truly innovative products. 

You will join a team developing state-of-the-art general-purpose robots designed to operate in human spaces and with human tools. It is designed to work alongside humans, mobilize to human spaces, and manipulate the world around it.

JOB SUMMARY 

At Apptronik AI Core Robotics team (AIRcore) team, you will build a new generation of world-class large behavior and reasoning models for multi-resolution decision making in Robotics and build production grade robotics solutions and deliver this to customers.

You will build a stack of fundamental perception and decision-making algorithms to create a new generation of Robot controllers and build production grade robotics solutions and deliver this to production. Implement core algorithmic infrastructure in a task-agnostic manner and conduct research to build perception and controllers for high-dimensional physical systems such as full-body humanoids and bimanual dexterous hands.

ESSENTIAL DUTIES AND RESPONSIBILITIES or KEY ACCOUNTABILITIES 

Solve essential problems to train world simulators at massive scale, develop metrics and scaling laws for physical intelligence, curate and annotate training data, enable real-time interactive generation, and study integration of world models with multimodal language models. Embrace the bitter lesson and seek simple methods that scale, with emphasis on strong systems and infrastructure.

Systems for training multimodal transformers at massive scale.

• Infrastructure for large-scale video data pipelines and annotation.
• Inference optimization and distillation for real-time generation.
• Methods for native multimodal generation in language models.
• Methods for ultra-long-context transformers.
• Quantitative evals for physical accuracy and intelligence.
• Scaling law science for video pretraining.

TECH STACK 

• Python
• Pytorch, JAX and XLA
• CUDA (C++ and Triton)

EDUCATION and/or EXPERIENCE 

• Expertise in Reinforcement Learning, Imitation Learning and Inverse RL, specifically for continuous control.
• Experience building training codebases for large-scale distributed ML models
• Experience with simulation engines such as Mujoco, IsaacLab, Drake or similar.
• Experience  ML Workflows and optimizing efficiency of distributed training systems and/or inference systems.
• Background in robotics is desired but not necessary. 
• Master's/ PhD degree in Computer Science, Machine Learning, Engineering, Robotics, or a related field. 
• Demonstrated track record. The portfolio can consist of: code, technical writing (white papers or blog posts), peer reviewed papers in Tier-1 Machine Learning, Robotics and Computer Vision
• 2 Years of relevant full-time experience post-PhD or 4 years of experience otherwise

PHYSICAL REQUIREMENTS  

• Prolonged periods of sitting at a desk and working on a computer
• Must be able to lift 15 pounds at times
• Vision to read printed materials and a computer screen
• Hearing and speech to communicate 

*This is a direct hire.  Please, no outside Agency solicitations. 

Apptronik provides equal employment opportunities to all employees and applicants for employment and prohibits discrimination and harassment of any type without regard to race, color, religion, age, sex, national origin, disability status, genetics, protected veteran status, sexual orientation, gender identity or expression, or any other characteristic protected by federal, state or local laws.