Founding Robot Learning Engineer

Haptica Robotics•San Francisco, CA

23h

About The Position

Haptica is seeking a Robot Learning Engineer to own the company’s robot-learning function. The role will establish in-house training capabilities to measure the effectiveness of Haptica’s tactile sensing products. This includes building training and data-collection pipelines, stress-testing sensors, deciding what data to collect and how, evaluating policies on a robot, and helping customers integrate Haptica data into their systems. The engineer will have substantial autonomy and will build on existing resources, including a Franka FR3 arm, basic glove-based hand tracking, and initial work on training policies. Initial goals are to build an internal data-collection pipeline with the hardware team, drive and publish research establishing the role of tactile data in learning, and inform how customers use Haptica data in their stacks. The role involves developing data-collection, training, and inference pipelines on off-the-shelf hardware integrated with Haptica; training imitation-learning policies with and without tactile data; creating benchmarks and evaluation methods; extending collection to egocentric glove data; and influencing sensing hardware decisions. The engineer will also track developments in robot learning, support customer integrations, and share findings through demos, writeups, and papers with academic collaborators.

Requirements

MS or PhD in robotics, computer science, electrical engineering, or a related field, or equivalent industry experience.
Extensive experience with production-level software and machine-learning engineering best practices.
Experience with modern deep-learning frameworks such as PyTorch.
Strong written and verbal communication skills, including the ability to explain complex topics clearly to teammates, customers, and investors.
Fundamental knowledge of running imitation learning, such as diffusion policy, ACT, or similar methods, from data collection through training, deployment, and debugging.
Comfort working with hardware, including cameras, IMUs, calibration, and time synchronization across sensors.
Intuition for hardware decisions that matter.
Comfort with ambiguity and enthusiasm for building from scratch, defining constraints, and partnering closely with internal and external teams.

Nice To Haves

Experience with dexterous manipulation, multi-fingered hands, or retargeting.
Relevant tactile-sensing research or experience, and a passion for tactile sensing.
Experience taking egocentric or wearable data through to a trained policy.
A strong publication record in a relevant field.
Prior work on humanoids or highly dexterous robotic platforms.

Responsibilities

Build data-collection, training, and inference pipelines on off-the-shelf hardware integrated with Haptica.
Train imitation-learning policies, including ACT and diffusion policy, with and without tactile data on tasks involving shear, occluded grasps, and delicate objects.
Define evaluation methods and benchmark tasks using publicly researched setups, metrics, and tooling to quantify the impact and task-to-task variance of tactile data.
Extend data collection from teleoperation to egocentric glove data.
Integrate hand tracking, test retargeting from human hands to grippers and multi-fingered hands, and run experiments to determine how many fingers to track.
Influence hardware decisions based on learning requirements, including sampling rate, resolution, synchronization, drift, and calibration.
Keep up with leading robot-learning practices and help integrate Haptica into customers’ data-collection and teleoperation setups.
Share findings through demo videos, writeups, and papers with academic collaborators.
Work alongside the hardware team to build the internal data-collection pipeline.
Drive and publish research establishing the role of tactile data in learning and inform the roadmap for customer use of Haptica data.