New opportunity

Robotics Engineer - Singapore: AI Specialist

General Robotics · Singapore

About this role

General Robotics is seeking a Robotics Engineer - AI Specialist to join their Singapore team, contributing to the development of an intelligence grid for physical AI. The role focuses on advancing state-of-the-art robotics and foundation models, applying AI to systems ranging from robotic arms to humanoids. Responsibilities include designing methods and infrastructure for foundation models, developing VLA models for robotics applications, and creating data pipelines for large-scale training. You will work with high-fidelity simulation platforms for synthetic data generation and collaborate on experimental design, model evaluation, and platform development for the GRID system. Minimum Qualifications: Undergraduate or Master’s degree (completed or in progress) in Robotics, Computer Science, or a related field; research experience in machine learning, robotics, and computer vision; experience developing robotics algorithms or ML models at scale; and proficiency in Python/C++ with deep learning frameworks like PyTorch or Jax. Candidates must have current work authorization in Singapore. Desired Qualifications: Master's or PhD in a relevant technical field; direct experience in robotics, CV, or ML research; first-author publications at major AI/robotics conferences (e.g., NeurIPS, CVPR, ICRA); and experience with simulation platforms (AirSim, CARLA, Isaac Sim, Mujoco, Maniskill) or modern computer vision sensors and algorithms. The position offers a base salary range of 82,000 SGD to 220,000 SGD, commensurate with skills and experience.

For full details about this role, click “Apply on company site” to view the employer’s original posting.

Skills for this role

RoboticsMachine LearningMultimodal Machine LearningComputer VisionLLMsVLMsVLAsDiffusion ModelsFlow MatchingReinforcement LearningImitation LearningPythonC++PyTorchJaxSLAMNavigationSensor FusionLIDAR3D Point Clouds

Checking your profile…

Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.

Understand the role

What matters most

Development of Vision-Language-Action (VLA) Models
The core of the role involves adapting foundation models for physical robotics. You must demonstrate the ability to bridge high-level AI reasoning with low-level robotic control, specifically for platforms like robotic arms and humanoids.
Scalable Data Pipeline Engineering
You are expected to design infrastructure for large-scale training. This requires proficiency in creating synthetic data pipelines using high-fidelity simulators and managing real-world data collection from teleoperation or motion capture.
Systems-Aware Model Design
The role requires balancing advanced research with practical engineering constraints. You must show you can factor in real-time inference, sensor fusion, and hardware limitations when selecting or designing machine learning architectures.

Likely challenges to think through

Bridging Research and Production
The challenge lies in translating state-of-the-art research into the GRID platform's scalable features. Prepare by highlighting examples where you successfully moved a model from a research prototype to a robust, reusable engineering component.
Cross-Site Collaboration and Synchronization
As the Singapore team functions as an extended research facility for the US headquarters, you will face the challenge of maintaining alignment on research goals across time zones. Prepare to discuss your experience in remote collaboration and maintaining technical documentation for distributed teams.
Simulation-to-Real Gap
Relying on high-fidelity simulators like Isaac Sim or Mujoco for training introduces the risk of domain shift. Prepare to discuss strategies for validating synthetic data and ensuring that models trained in simulation perform reliably on physical hardware.

Practice your answers

Technical practice

Vision-Language-Action (VLA) Model Architecture

What to coverExplain how you would adapt a foundation model for robotic control. Discuss the trade-offs between end-to-end VLA approaches versus modular pipelines, and how you handle the integration of visual inputs with action space outputs.

Synthetic Data Generation in Simulation

What to coverDescribe your experience using platforms like Isaac Sim or Mujoco for synthetic data generation. Address how you bridge the 'sim-to-real' gap and ensure that data generated in high-fidelity simulations effectively trains models for physical robot deployment.

Large-Scale Training Pipelines

What to coverDetail your approach to designing data loading and training pipelines for foundation models. Focus on how you manage data throughput, handle multi-modal datasets, and optimize training efficiency using frameworks like PyTorch or Jax.

Sensor Fusion and 3D Perception

What to coverDiscuss the challenges of integrating data from diverse sensors such as RGB-D cameras and LIDAR. Explain how you process 3D point clouds or meshes for tasks like object detection, segmentation, or SLAM in real-time robotics applications.

Reinforcement and Imitation Learning

What to coverCompare the use of Reinforcement Learning (RL) versus Imitation Learning (IL) for training robotic arms or humanoids. Suggestion: Prepare to discuss how you balance exploration in RL with the data efficiency of IL when working with limited real-world demonstrations.

Systems Considerations for Model Inference

What to coverExplain how you factor system constraints—such as latency, power, and compute availability—into your choice of model architecture. Discuss how you optimize inference for deployment on physical hardware.

Diffusion Models and Flow Matching

What to coverDescribe the application of generative models like diffusion or flow matching in a robotics context. Explain how these models can be used for trajectory generation or policy learning compared to traditional control methods.

Experimental Design and Evaluation

What to coverOutline your process for designing experiments to evaluate a new robotics algorithm. Suggestion: Focus on how you define success metrics, ensure reproducibility, and organize results to inform iterative model development.

Behavioural practice

Bridging Research and Engineering

What to coverDescribe a situation where you had to translate a complex research concept into a practical, deployable engineering solution. Focus on how you balanced theoretical innovation with the constraints of a production environment.

Cross-Site Collaboration

What to coverThe role involves working closely with a US-based team from Singapore. Provide an example of how you have effectively collaborated with remote teams to align on research goals and maintain project momentum across different time zones.

Handling Technical Ambiguity

What to coverTell me about a time you were tasked with a research goal that lacked a clear path forward. Explain how you defined the scope, identified necessary tools, and communicated your progress to stakeholders.

Data Collection Challenges

What to coverDescribe a project where you led a data collection campaign, whether via teleoperation, motion capture, or simulation. Discuss how you handled unexpected data quality issues and the steps you took to rectify them.

Sharpen your knowledge

Topics to prepare

Vision-Language-Action (VLA) Models
Understand the architecture and training objectives of VLA models that map visual inputs and language instructions to robotic actions. Exercise: Review recent literature on VLA architectures and explain how they differ from standard LLMs in a robotics context, specifically regarding action space discretization or continuous control.
High-Fidelity Simulation for Synthetic Data
Focus on the role of simulation platforms like Isaac Sim or Mujoco in generating synthetic data for large-scale training. Exercise: Design a hypothetical data pipeline that uses simulation to bridge the sim-to-real gap, focusing on how to randomize environmental parameters to improve model robustness.
Reinforcement and Imitation Learning
Study the application of RL and IL to physical robots. Exercise: Compare the trade-offs between imitation learning from teleoperation data versus reinforcement learning from scratch, considering sample efficiency and safety constraints in real-world deployment.

Skills to revise

Deep Learning Frameworks (PyTorch/Jax)
Demonstrate proficiency in implementing custom neural network layers or training loops. Exercise: Write a PyTorch or Jax script that implements a simple transformer-based policy network, focusing on efficient tensor operations and memory management for large-scale training.
Robotics Sensor Fusion and Perception
Develop skills in processing multi-modal sensor data (RGB-D, LIDAR) for 3D scene understanding. Exercise: Create a pipeline that fuses point cloud data with RGB images to perform object detection or segmentation, ensuring the output is suitable for downstream navigation or manipulation tasks.
C++ and Python for Robotics Systems
Ensure code is optimized for real-time inference and system integration. Exercise: Refactor a Python-based prototype into a C++ module to demonstrate an understanding of performance bottlenecks, memory management, and integration with standard robotics middleware.

Your preparation checklist

Prepare a technical portfolio of robotics research
Curate a portfolio highlighting your first-author publications or significant projects in machine learning, computer vision, or robotics. Focus on demonstrating your ability to scale models and your proficiency in Python, C++, and frameworks like PyTorch or Jax.
Develop a simulation-based project walkthrough
Prepare a detailed explanation of a project where you utilized high-fidelity simulation platforms such as Isaac Sim, Mujoco, or Maniskill. Be ready to discuss how you generated synthetic data and translated those results to real-world robotics applications.
Review VLA and foundation model architectures
Deepen your understanding of Vision-Language-Action (VLA) models and their application in physical AI. Be prepared to discuss the trade-offs between different model architectures and how you would design training pipelines for large-scale robotics data.
Formulate questions about the GRID platform
Prepare thoughtful questions regarding the technical challenges of the GRID system, such as how the team balances research goals with practical engineering constraints or how they manage cross-continental collaboration between the Singapore and US offices.
Synthesize your experience with sensor fusion and navigation
Draft concrete examples of how you have implemented sensor fusion (e.g., LIDAR, RGB-D) and SLAM algorithms. Connect these experiences to the role's requirement for building efficient robotics pipelines and model inference.

Availability last checked: 18 Oct 2026. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy

KEEP EXPLORING

Similar AI jobs

Related skills and specializations in Singapore. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…