Research Scientist / Engineer: Reinforcement Learning

← Back to Careers

Research Scientist / Engineer: Reinforcement Learning

Seoul Gangnam (On-site)

• Full-time

About The Role

Holiday Robotics builds FRIDAY, a dexterous industrial humanoid designed for repetitive and physically demanding work. We develop the full stack in-house—from hardware and firmware to control, simulation, learning, perception, and operational software—because reliable manipulation depends on every layer working together.

We are looking for reinforcement learning researchers and engineers who can turn learning into reliable, dexterous behavior on FRIDAY. You will train policies in our in-house RL stack and deploy them on the real robot. Much of the work is research: advancing the learning algorithms themselves, building benchmarks that let us systematically generate and evaluate new skills, finding reward formulations that generalize across many tasks so we depend less on hand-tuned per-task rewards, and developing methods that learn from only a small number of demonstrations, letting RL expand that data on its own.

Depending on your background, your work may center on RL algorithm research, benchmark design and skill generation, generalizable rewards, learning from limited demonstrations, contact-rich manipulation, or closing the sim-to-real gap on real hardware.

Key Areas

RL algorithm researchBenchmarks & skill generationGeneralizable rewardsDemonstration-efficient RLDexterous manipulationSim-to-real transfer

What You'll Do

• Advance the learning algorithms for dexterous, contact-rich control.
• Design environments and rewards that generalize across tasks, reducing reliance on hand-tuned per-task rewards.
• Develop methods that learn from only a small number of demonstrations, letting RL expand that data on its own.
• Build benchmarks that let us systematically generate, measure, and compare new robot skills.
• Close the sim-to-real gap and train policies with reproducible, well-measured pipelines.
• Deploy learned policies on FRIDAY and collaborate across control, simulation, perception, hardware, and data teams.

Required Qualifications

• BS, MS, or Ph.D. in Robotics, Computer Science, Electrical Engineering, Machine Learning, or equivalent practical experience.
• Strong foundation in reinforcement learning, optimization, and probability, with a solid understanding of deep learning.
• Hands-on experience training RL policies and diagnosing training dynamics.
• Strong Python engineering skills, familiarity with PyTorch, and the ability to debug across algorithms, simulation, and hardware.
• Ability to translate research ideas into measurable behavior on physical systems.

Preferred Qualifications

• Experience applying RL to manipulation, dexterous hands, locomotion, or whole-body control.
• Experience with large-scale and sample-efficient RL training.
• Experience bridging the sim-to-real gap through domain randomization, system identification, or learned residual models.
• Experience developing data-efficient methods that learn from a small number of demonstrations.
• Experience with contact-rich or bimanual manipulation.
• Publications, open-source contributions, competition results, patents, or shipped robotic systems demonstrating exceptional work.

What We Offer

• Competitive compensation based on experience, level, and technical impact.
• The tools, compute, robots, and equipment needed to do your best work.
• The opportunity to build, test, and deploy directly on FRIDAY and the Holiday Robotics full stack.
• Close collaboration with researchers and engineers across hardware, control, simulation, learning, perception, data, and operations.

Apply