Senior/Staff Machine Learning Engineer

As a Senior/Staff Machine Learning Engineer, you will be working on a myriad of challenges related to robot task and action planning

You will leverage techniques from machine learning to solve hard sequential decision problems that require reasoning about the physical world and its dynamics

You will also stay abreast of the latest progress in imitation learning, reinforcement learning, and other related fields in order to further develop Dexterity’s technology foundations in Physical AI

Additionally, you will be responsible for updating and scaling our current ML pipelines to cover more scenarios and improve accuracy

Dexterity’s robotic solutions integrate data from a multitude of sensors, including RGB cameras, depth sensors, force-torque sensors, encoders, system telemetry and human input

To better inform the planning algorithms, you may work on sensor fusion and state estimation techniques to leverage this multimodal sensory data

You will also work closely with the data platform, physics simulation, and robot operations teams to develop effective and efficient ways to improve the system’s internal world model

Dexterity has an expanding set of algorithmic challenges as we deploy new robotic applications, including areas such as:

Improving packing algorithms to build taller, denser, more stable structures with a wider variety of objects

Solving the logistics task of moving and sorting inventory throughout a warehouse

Building models that understand physics and geometry for both short- and long-horizon tasks

In addition to curating datasets and developing/improving machine learning models, you will be responsible for building data flywheels

Design and implement machine learning solutions across Dexterity’s robotics stack, including but not limited to perception, decision-making, action scoring, and predictive modeling

Own the full ML development cycle for these solutions: data curation, labeling, training, evaluation, deployment, and iteration

Build performant training and inference pipelines using PyTorch, with production-readiness and scalability in mind

Collaborate closely with robotics, data platform, and simulation teams to integrate ML into real-time, latency-sensitive robotic systems

Use profiling, monitoring, and experiments to optimize model performance and reliability

Ensure reproducibility, traceability, and modularity across training and serving pipelines

Maintain clean, production-quality code in Python (and C++ where required)

Help establish best practices for model versioning, dataset management, and ML operations- Ideally, you will bring data-driven productization experience and help the team broadly in qualifying, deploying and updating models

Experience with cloud-based infrastructure (AWS, GCP, Azure) and containerized environments (Docker)

Strong engineering background with experience profiling, debugging, and optimizing model and pipeline performance

Proven ability to design and maintain reliable systems, from model training to field deployment

Ability to work fluently across ML tasks, e.g., classification, regression, ranking, segmentation, and structured prediction

Familiarity with Linux, Git, CI/CD) and software development best practices (unit/acceptance/integration testing, code reviews)

Degree in Computer Science, Electrical Engineering, or Mathematics 5+ years of industry experience applying machine learning to real-world, production systemsStrong Python skills and deep experience with PyTorch

Prior experience in robotics, autonomous systems, or real-time ML applications

Exposure to multimodal data (e.g., RGBD, force-torque, pose estimates, telemetry)

Experience deploying models using serving stacks like NVIDIA Triton, TorchServe, or custom low-latency frameworks

Experience with Kubernetes, Ray or other distributed training and inference systems

Background in computer vision, geometric learning, or time-series modeling

Previous startup experience or experience in fast-paced, cross-disciplinary environments