D

Dyna Robotics

General-purpose robots with proprietary foundation models

Research Engineer

AI ResearchUpdated on Nov 13
Full-timeRedwood City, CA, USANo salary listed

About the Job

Requirements

  • Genuinely care about embodied AI, follow robot learning research, have opinions about vision-language-action models or world models, or have built something that moves.
  • Have 7 or more years of engineering experience, including leading projects in high-performance computing or machine-learning infrastructure.
  • Know PyTorch and distributed training deeply, including DeepSpeed, Accelerate, Fully Sharded Data Parallel, mixed precision, and gradient accumulation.
  • Have hands-on experience with cloud GPU environments such as Google Cloud Platform or Amazon Web Services, and Kubernetes.
  • Understand distributed systems at a low level, including race conditions, memory management, NVIDIA Collective Communications Library, and inter-node communication.
  • Design, build, and operate systems end to end to unblock fast-moving research.

Responsibilities

  • Scale distributed training across a multi-cloud GPU fleet using sharding, activation checkpointing, and memory optimization such as Fully Sharded Data Parallel and ZeRO.
  • Build a research codebase and scheduling system using Kubernetes or SLURM for fast iteration, automatic retries, and failure recovery.
  • Design high-throughput pipelines for terabytes of multimodal robot data so data loaders do not starve GPUs.
  • Ship low-latency inference for real-time robot control using quantization, distillation, and compilation with TensorRT or Triton.
  • Profile GPU utilization, I/O bottlenecks, and memory fragmentation to maximize node performance.

Desired Qualifications

  • Robotics data formats such as MCAP or Protocol Buffers, or multimodal models such as vision-language-action models.
  • Custom kernels using Triton, compilers, or runtime optimization.
  • Experience as a founding or early infrastructure hire.
  • Side projects involving robots, simulators, or hardware.

About the Company

Dyna Robotics

General-purpose robots with proprietary foundation models

Dyna Robotics designs and deploys general-purpose robots that are powered by proprietary foundation models. Their DYNA-1 model enables embodied-AI robots to operate autonomously across real-world commercial environments and continuously improve through experience. The robots work by running a single base model that generalizes to tasks in different settings—hotels, restaurants, laundromats, and gyms—allowing multi-shift, 24/7 operation with high task success. Compared with others, Dyna focuses on broad applicability and self-improvement of their embodied AI in diverse commercial contexts, backed by a strong founder track record and large financing rounds from major investors to scale research, engineering, and production-ready deployments. The company aims to expand research and development and extend production-ready deployments across more commercial environments.

Company size
51-200
Company stage
Series A
Total funding
$143.5M
Headquarters
Redwood City, CA
Founded
2024

Recently Posted Jobs

Explore more opportunities at Dyna Robotics

See more jobs