Founding Engineer — Robotics Data Infrastructure

Role focus

Neural Motion is building infrastructure for robotics learning: a universal data pipeline and cross-embodiment representation layer that unifies real-world logs, simulation, and multimodal datasets. The platform powers public datasets and tooling for researchers, enterprise data pipelines, and cross-embodiment learning at scale. Early engineers own core systems from large-scale ingestion to embodiment-aware transformations so knowledge learned on one robot can transfer across many.

Responsibilities

• Design and build high-throughput data pipelines for ingesting, processing, and standardizing robotics datasets.
• Architect distributed systems and microservices for robotics data processing and dataset infrastructure.
• Develop the data compiler layer that standardizes raw logs into a unified representation.
• Build cross-embodiment transformation pipelines (retargeting, normalization, alignment).
• Integrate multimodal augmentation models (vision, language, SLAM, simulation).
• Enable real ↔ sim pipelines and unified evaluation frameworks.
• Build tooling for dataset ingestion & validation, annotation and enrichment, and dataset versioning and reproducibility.
• Ship public dataset surfaces (APIs, SDKs, data loaders), internal enterprise sourcing pipelines, and interfaces for model training and evaluation.
• Work with founders on architecture and direction; integrate research outputs from robotics learning experiments into the platform.

Minimum qualifications

• One role, two strong profiles we actively consider (often combined in one person):
• (A) Infrastructure / distributed systems — large-scale data systems (TB–PB), streaming pipelines, microservices; tools such as Kafka or Pulsar, Temporal or Airflow, AWS or GCP (e.g.
• S3, SQS, Lambda, ECS/EKS); robust fault-tolerant pipelines; strong backend in Python, Go, or similar.
• (B) Robotics / robot learning — embodied AI or manipulation; familiarity with VLA or world models, imitation or RL, robotics dataset design; kinematics (FK/IK), retargeting, coordinate frames and calibration; ROS/URDF and simulation (e.g.
• Isaac Gym, MuJoCo).
• Mission-driven; systems thinking; cares about data quality, structure, and scale; comfortable in ambiguous, fast-moving early-stage environments.

Preferred qualifications

• Experience spanning both infrastructure and robotics, or deep collaboration across both.
• Large multimodal datasets (video, sensor logs).
• Dataset platforms (HuggingFace, TFDS, RLDS).
• Internal tools for ML/data teams.
• Simulation ↔ real transfer.
• Startup or zero-to-one experience.

Compensation

Base salary approximately $150,000–$220,000 USD; roughly ~1% early-employee equity (experience-dependent).