Robotics Data Pipeline Engineer – Multimodal Data

Persona AI Inc•Houston, TX

1d•Onsite

About The Position

As a Data Pipeline Engineer, you will architect and scale the data infrastructure that feeds our foundation models. Your primary mission is to extract, augment, and align human dexterous manipulation data from massive complex, multi-sensor and egocentric video datasets. Crucially, you will build advanced post-processing algorithms to perform deep force analysis and infer hidden states from raw data—such as processing direct force-torque outputs to quantify grasp dynamics, estimating contact forces from visual cues, extrapolating heavily occluded hand positions, or deriving 3D geometry from 2D frames. You will use spatial, temporal, and cross-modal data augmentation to multiply the value of every minute of data our teleoperation team collects.

Requirements

B.S., M.S., or Ph.D. in Computer Science, Data Engineering, Machine Learning, Robotics, or a related field.
Deep expertise in Python and extensive experience with PyTorch, specifically in handling custom dataloaders for multimodal datasets.
Experience analyzing and processing complex time-series data from force-torque (F/T) sensors, load cells, or tactile arrays, ensuring pristine alignment with visual frames.
Mastery of video processing pipelines and libraries (OpenCV, FFmpeg, Decord) and managing the I/O bottlenecks of terabyte-scale video datasets.
Hands-on experience with 3D hand tracking, human pose estimation (e.g., MediaPipe), and spatial geometry calculations.
Strong understanding of modern imitation learning paradigms, VLA architectures, and frameworks focused on human-to-robot transfer (e.g., EgoScale, EgoMimic, or OpenVLA).
Proven ability to implement programmatic and generative data augmentation techniques for computer vision and time-series data.

Nice To Haves

Experience with NVIDIA’s robotic software stack (Isaac, Cosmos, or components of the GR00T framework).
Familiarity with distributed data processing systems (Ray, Apache Spark) for cluster computing.
Background in generating or utilizing synthetic robotic data via simulation (Omniverse, MuJoCo).
Experience integrating spatial awareness or tactile data representations (e.g., Fourier encoding) into visual pipelines.

Responsibilities

Architect highly efficient, scalable pipelines to ingest, decode, and synchronously process thousands of hours of high-resolution egocentric video alongside rich sensor streams (IMUs, force-torque sensors, tactile pads, and joint proprioception).
Develop sophisticated post-processing algorithms to analyze force interactions and infer unobservable or missing states from raw data. This includes calibrating and cleaning direct force-aware data collections, estimating contact forces from object deformation, tracking occluded objects during complex manipulation, or applying inverse kinematics to fill in missing joint trajectories.
Develop algorithms to translate 3D human hand tracking, wrist motion, and pose estimation into the specific 6DoF/joint-space coordinates of our humanoid’s end-effectors, relying on sensor fusion to ensure absolute precision.
Implement robust data augmentation strategies (spatial transformations, temporal scaling, synthetic viewpoints, and sensor noise injection) to expand expert trajectories and improve the robustness of our learning models.
Work closely with the Hardware Teleoperation Team (UMI & Console operators) to perfectly align human-robot play-data (haptics, force profiles, video, audio, telemetry) with large-scale pre-training datasets.