Robotics Research Engineer

About Abaka AI

Abaka AI is built on one mission: to be the world's most trusted data partner for AI companies. More than 1,000 industry leaders across Generative AI, Embodied AI, and Automotive AI rely on us to power their data pipelines. With our headquarters in Silicon Valley and teams in Paris, Singapore, and Tokyo, we support global partners with fast, reliable, and scalable data solutions.

Our offerings include a diverse catalog of off-the-shelf datasets (image, video, multimodal, reasoning, 3D, and beyond) as well as comprehensive data collection and annotation services. Whether teams need raw data, curated datasets, or full-cycle data engineering, Abaka AI provides the foundation for building high-performance AI systems.

We are looking for a Robotics Research Engineer to join Abaka Robotics Lab, our in-house research group focused on building the data foundation for physical AI.

We believe the next wave of AI will act in the physical world—and that its greatest bottleneck will be data. Unlike language and vision, there is no internet-scale corpus of robot experience readily available. That data must be captured, synthesized, labeled, and carefully curated. Abaka already works closely with several leading physical AI labs to help build this foundation.

This is a hands-on, research-driven role combining robot learning, data production, and publishable research. You will train and evaluate policies on both simulated and real-world robotic systems, use experimental results to understand which data matters, and design scalable pipelines that turn raw sensor data into high-quality training datasets with minimal human intervention.

You will work with hardware including 6-DoF collaborative robot arms and an ALOHA-style bimanual platform with full teleoperation. Your findings will directly inform the datasets we build for our partners and contribute to publications at leading robotics, computer vision, and machine learning venues.

Scope 1:
Robot Learning Research

Train and evaluate robot policies (imitation learning, diffusion policies, VLA, RL fine-tuning) on our data and public baselines, in simulation and on real robots.

Turn evaluation results into a quality signal for the data: which data helped, which did not, and why.

Track the state of the art: reproduce the papers that matter, run current methods on our hardware, and develop findings that hold up into publications at top robotics, vision, or ML venues.

Maintain the training and evaluation infrastructure: training runs, evaluation harnesses, and the simulation or world-model environments used ahead of hardware tests.

Scope 2:
Data Pipeline:
Synthesis, Annotation, Curation

Design pipelines that turn raw capture (egocentric human video, robot logs, teleoperation, 3D scans) into training data with minimal human labeling: synthesize what was not captured, auto-label with models and geometry, and curate what goes into training.

Track which data decisions changed policy performance.

MS or PhD in Robotics, Computer Science, Machine Learning, or a related field.

1 - 3 years of hands-on experience with real robot hardware, through research, an internship, or a project.

Strong machine learning fundamentals: optimization, generalization, evaluation methodology, and end-to-end model training.

Strong robotics fundamentals: coordinate frames, kinematics, basic control.

Strong 3D vision fundamentals: camera models, multi-view geometry, point clouds.

Rigor in evaluation: claims about a policy or a dataset are backed by measurements.

Publications or workshop papers at top-tier robotics (CoRL, RSS, ICRA, IROS), vision (CVPR, ICCV, ECCV, 3DV), or ML (NeurIPS, ICML, ICLR) venues. A widely used open-source repository or a dataset adopted by other teams is valued equally.

Experience with imitation learning or VLA codebases such as LeRobot, ACT, Diffusion Policy, OpenVLA, or pi0.

Experience deploying a learned policy on a real robot and running a rigorous evaluation.

Practical 3D reconstruction experience: COLMAP, SLAM, 3

DGS/NeRF, point-cloud registration.

World models or video generative models, particularly as simulators or data generators.

RL fine-tuning of learned policies, or model-based RL.

Hand and object pose estimation, hand-object interaction, or human-to-robot motion retargeting.

Depth in a simulator (Mu Jo Co , Isaac Sim/Lab, Genesis, SAPIEN), including environment and asset authoring.

Domain randomization, procedural…