Robot Learning Engineer

Cubiq Recruitment; New Jersey; On-site; USD 180,000 - 280,000; Full time

Job summary

Cubiq Recruitment's client, an early-stage robotics startup building AI-powered automation for the garment industry, seeks a Founding Robot Learning Engineer to build the core technology from the ground up.

This hands-on role spans robot learning research, foundation model development and real-world robotics, with ownership over the direction of the technology. You will prototype, experiment, and deploy policies on real hardware.

Qualifications

• PhD or equivalent in robot learning with research or hands-on systems.
• Experience training large-scale VLAs, VLMs, diffusion/world models.
• Proven track record shipping policies to real hardware.
• Willingness to own direction in an early-stage startup.

Responsibilities

• Research and decompose challenging manipulation tasks.
• Develop core model architectures, training pipelines, and evaluation systems.
• Collect data and train policies on real robot platforms.
• Build simulation environments and maintain experiment infra.

Skills

Robot learning

Large-scale models

Hands-on systems

Policy deployment

Education

PhD in robotics

Tools

Simulation environments

Data pipelines

RL pipelines

Job description

We’re working with an early-stage robotics startup building AI-powered automation for the garment industry, backed by Jeff Dean, Pieter Abbeel and senior leaders from OpenAI.

They’re looking for a Founding Robot Learning Engineer to help build the core technology from the ground up.

This is a hands-on role spanning robot learning research, foundation model development and real-world robotics.

What you’ll be doing

• Research & problem decomposition: Break down challenging manipulation problems into testable hypotheses and resolve them through rapid experimentation.
• Core model development: Own architectures, training pipelines and evaluation systems for dexterous automation.
• Real-world experimentation: Collect data, train policies and evaluate them directly on physical robot platforms.
• Experiment infrastructure: Build and maintain simulation environments and commercial robot platforms to keep experiments running reliably.
• Push the boundaries of VLA, VLM, diffusion and world models for real-world robotic applications.
• Take models from pretraining and fine-tuning through to RL-based post-training and deployment.

Ideal Background

• PhD or equivalent depth in robot learning, either through academic research or hands-on systems building.
• Strong experience training large-scale VLAs, VLMs, diffusion models or world models.
• Experience taking models through the full lifecycle, from pretraining and fine-tuning to RL-based post-training.
• A track record of shipping learned policies on real hardware, including data collection, training and evaluation.
• Excited by the opportunity to work at an early-stage company and have significant ownership over the direction of the technology.