Physical AI Engineer / Generative AI Engineer

Job Title: Physical AI Engineer / Generative AI Engineer

Company: Muks Robotics AI Pvt. Ltd.

Website: https://muksrobotics.com/

Location: Baner, Pune

Experience: 2-3 years

About the Role :

As a Physical AI Engineer / Generative AI Engineer, you will be responsible for integrating LLMs, VLMs, and VLA models directly onto our humanoid robots, enabling real-time perception, reasoning, control, and manipulation. This is a highly hands-on role involving robots, sensors, actuators, embedded compute, and full-stack AI pipelines.

Key Responsibilities:

1. AI Integration & Deployment

● Deploy and adapt LLM/VLM/VLA models on edge platforms like Jetson Orin, Nvidia GPUs and ARM-based hardware.
● Connect AI outputs to robot software pipelines for task execution and interaction.
● Optimize models for speed, reliability, and on-device inference.

2. Perception & Sensing

● Build vision and multimodal perception modules for scene understanding, grounding, and pose estimation.
● Combine data from cameras, force/torque sensors, tactile sensors, and robot joints.
● Implement lightweight memory/continuity modules for stable task performance.

3. Control & Execution

● Link high-level AI decisions with motion controllers for manipulation, locomotion, and coordinated actions.
● Validate behaviors generated by AI models and refine parameters for smooth, safe operation.
● Handle timing, safety checks, and real-time control loops.

4. Experimentation & Hardware Work

● Run hands-on tests on humanoid robots for tasks like pick/place, handovers, and tool usage.
● Work on low-latency communication between hardware components (ROS/ROS2, CAN, serial protocols).
● Diagnose issues across sensors, actuators, drivers, and communication interfaces.
● Set up logging, telemetry, and evaluation tools for systematic testing.

5. Data Workflows

● Capture multimodal datasets from real robot trials.
● Build simple pipelines for preprocessing, organization, and annotation.
● Support demonstration-based learning and teleoperation data collection.

6. Safety & Reliability

● Implement guardrails, fallback behaviors, and monitoring systems to ensure safe AI-driven actions.

Required Skills

● Bachelor’s or a Master’s degree in Computer Science, AI/ML or related field.
● Strong experience integrating LLMs/VLMs/VLA models into robotic systems (grounding, reasoning, action planning).
● Solid foundation in deep learning tools: PyTorch, ONNX, TensorRT.
● Experience with ROS/ROS2 and real-time robot bring-up.
● Proficiency in Python and C++.
● Hands-on experience with robotics platforms: manipulators, grippers, mobile bases, or humanoids.
● Good understanding of kinematics, dynamics, control loops, and robotic safety.
● Experience working with Jetson/embedded inference systems.

Preferred (Not mandatory):

● Experience with generalist robotic models (RT-X, GR00T, OpenVLA, Octo, VIDIA-like architectures, or internal equivalents).
● Experience building teleoperation, imitation learning, or demonstration-based skill pipelines.
● Knowledge of SLAM, 3D reconstruction, depth fusion, or scene graph generation.
● Familiarity with memory systems, vector databases, or trajectory libraries for robotic decision-making.
● Exposure to whole-body control: MPC, QP solvers, whole-body IK.
● Contributions to open-source robotics or AI frameworks.

What We Offer:

● Opportunity to work on futuristic robotics technologies.
● A collaborative and innovative work environment.
● Growth opportunities in a rapidly expanding tech company.
● Competitive salary and performance-based incentives.

Job Type: Full-time

Work Location: In person