Senior Robotics Software Engineer, Robot Learning and Manipulation

About Us:

Phridom AI - Ability Robotics is an early-stage robotics and AIcompany building the core systems behind embodied intelligence. Ourtechnologies help robots understand the physical world, learn intelligentbehaviours, and operate reliably in real environments. We’re still in stealth,but we’re already working with a select group of partners to validate oursystems in high-impact robotics applications. If you're excited to buildfoundational technologies for the next generation of robotics and wantmeaningful ownership over hard technical problems, we'd love to hear from you.

Role Overview:

As an Embodied AI Research Engineer, you'll develop the learningsystems that enable robots to understand, reason, and act in the physicalworld. You'll work at the intersection of Vision-Language-Action models,reinforcement learning, and real-world robotics, building end-to-endintelligence that translates perception into robust robot behavior. From modeldesign and training to deployment on physical robots, your work will directlyshape the capabilities of next-generation embodied AI systems operating in realenvironments.

What you'll work on

1.                         Develop and deploy intelligent algorithms for real-world roboticsystems, with a focus on Vision-Language-Action (VLA) models and reinforcementlearning (RL);

2.                          Design and implementend-to-end learning systems from perception to decision-making and control,improving generalization and task success rates in complex real-worldenvironments;

3.                          Lead or contribute to thetraining, evaluation, and deployment of VLA models in robotic applications,including data pipeline design, model architecture, and inference optimization;

4.                          Develop policy learningalgorithms using reinforcement learning (RL), imitation learning (IL), orrelated methods, ensuring stable execution on real robotic platforms;

5.                          Drive Sim-to-Real transfer,addressing challenges such as domain gaps, policy stability, and safety inreal-world deployment;

6.                          Collaborate closely withperception, simulation, and system engineering teams to close the loop betweendata, models, and system performance;

7.                          Build reproducible trainingand evaluation pipelines, and continuously improve robustness, stability, andefficiency in real-world deployments;

8.                          Stay up to date withstate-of-the-art research (e.g., VLA, Embodied AI, Foundation Models) andtranslate advances into practical robotic applications.

Technical Requirements

Must Have

1.                         PhD in Computer Science, Robotics, Artificial Intelligence, or arelated field;

2.                          Strong foundation in machinelearning and deep learning, with deep expertise in reinforcement learning (RL),imitation learning (IL), or related areas;

3.                          Experience inVision-Language-Action (VLA) or Embodied AI research or engineering;

4.                          Proficiency in at least onemajor deep learning framework (e.g., PyTorch or TensorFlow), with strongimplementation skills;

5.                          Familiarity with key roboticslearning challenges, such as policy learning, generalization, and Sim-to-Realtransfer;

6.                          Experience deployingalgorithms on real robotic systems, or a strong interest in bridging researchand production;

7.                          Strong problem decompositionskills and system-level thinking, with the ability to drive algorithmdeployment in complex systems;

8.                          Strong communication andcross-functional collaboration skills.

Nice to Have

1.                         Publications in top-tier conferences (e.g., NeurIPS, ICML, ICLR,CVPR, CoRL, RSS);

2.                          Hands-on experience deployingalgorithms on real robotic platforms (e.g., manipulation or navigation);

3.                          Experience training oroptimizing large-scale models (e.g., foundation models or VLA models);

4.                          Experience deployingreinforcement learning in real-world systems (beyond simulation-only work);

5.                          Experience with large-scaledata pipelines, dataset construction, or automated labeling;

6.                          Experience with multimodallearning systems (vision-language-action);

Your Impact

1.                         You'll help define and deploy the intelligence stack behindnext-generation embodied robots, enabling them to perceive, reason, and act incomplex real-world environments.

2.                         Your work will bridge cutting-edge AI research and real-world roboticdeployment, transforming state-of-the-art models into reliable roboticcapabilities.

3.                         You'll advance the practical application of Vision-Language-Actionmodels, reinforcement learning, and foundation models in real robotic systems.

4.                         You'll play a key role in building scalable learning systems thatimprove robot generalization, adaptability, and autonomy across a wide range oftasks and environments.

Why Join Us

Join us at an early stage and help build the core softwareinfrastructure behind humanoid and embodied robotic systems heading towardreal-world deployment. We move fast, ship often, and give engineers meaningfulownership—so you'll learn quickly, tackle high-impact challenges, and see yourwork running on real robots outside the lab. With active business across Chinaand beyond, you'll gain unique exposure to global robotics ecosystems,customers, and partners as the company scales.

Location

This role is based on-site in Singapore, working closely with theengineering team and robotic systems.

Occasional international travel to China and other regions may berequired for training, customer engagements, deployments, and partnercollaborations.

How to Apply

Send your resume and a brief introduction outlining your relevantexperience and interest in the role to [email protected].

We'd also love to see examples of simulation environments, roboticsprojects, synthetic data pipelines, or other work you're proud of.