Scientist/Researcher | Robotics | ML & Computer vision

Randstad Singapore; Singapore; On-site; SGD 90,000 - 150,000; Full time

Job summary

Randstad Singapore is seeking a robotics ML researcher to develop perception and learning algorithms for real-world robotic systems. You will design representation models for whole-body control and predictive world models, evaluating them on physical hardware and deploying robust behaviors beyond lab environments.

You will stay current with AI literature, contribute to open-source projects, and collaborate with cross-functional teams to ensure scalable solutions in a fast-growing robotics domain.

Qualifications

• Strong foundations in computer vision with real-world visual data.
• Experience in representation learning and self-supervised training methods.
• End-to-end model training with data pipelines, debugging and evaluations.
• Direct experience with sim-to-real transfer onto physical hardware.
• Proficiency in Python or C++, and familiarity with robotic simulation environments.
• Broad ML awareness including large language models (LLMs).
• Knowledge of imitation learning or embedded systems control is advantageous.

Responsibilities

• Develop perception and learning algorithms for robotic systems.
• Design representation models for whole-body control and predictive world models.
• Evaluate models on physical hardware and deploy robust behaviors outside the lab.
• Stay current with AI literature and contribute to open-source initiatives.
• Collaborate with cross-functional teams to ensure reliable deployment.

Skills

Computer vision

Representation learning

Self-supervised learning

Model training

Sim-to-real transfer

Python

C++

Robotics simulation

LLMs

Imitation learning

Job description

Exciting time to be in a fast growing company

Robotics at the forefront!

about the company

Our client is an applied R&D laboratory operating within the robotics and artificial intelligence industry. They focus on developing advanced open-source robotic platforms and building the comprehensive software stacks required to power them. Dedicated to making scalable automated labor economically viable, the organization builds rapidly across hardware, autonomy, and simulation domains to deploy solutions in the real world.

about the role

You will be responsible for developing the perception and learning algorithms that enable robotic systems to function effectively in physical environments. Your primary tasks will involve designing representation models for whole-body control and advancing predictive world model research. You will evaluate these models directly on physical hardware and collaborate closely with cross-functional teams to deploy robust behaviors outside of a laboratory setting. The position also requires staying current with broader AI literature and contributing to open-source initiatives.

skills and experience

• Strong foundations in computer vision with a track record of handling uncurated, real-world visual data.
• Hands-on expertise in representation learning and self-supervised training methods.
• Proven end-to-end experience in model training, including managing data pipelines, debugging, and running comprehensive evaluations.
• Direct experience with sim-to-real transfer, actively migrating models from simulation onto physical hardware.
• High proficiency in Python or C++, alongside strong familiarity with robotic simulation environments.
• A broad understanding of the wider machine learning landscape, including large language models (LLMs).
• Familiarity with imitation learning, real-time control on embedded systems, or contact-rich manipulation is highly advantageous.