Full-Stack Robotics Engineer (Embodied AI)

# Full-Stack Robotics Engineer (Embodied AI)

## About Axonex Robotics

Axonex Robotics is an innovative technology company specializing in robotics and AI solutions, delivering intelligent products across security patrol, education, and companion domains. With a strong technical team and established product portfolio, we're advancing embodied intelligence to create robots that truly understand the world.

Our Tech Stack:

- Robotics: ROS/ROS2, Motion Control, SLAM, Navigation

- AI: VLA/VLM, YOLO, Foundation Pose, Imitation Learning

- Systems: NVIDIA Jetson, Docker, Edge AI

- Data: Robot telemetry, Replay system, Synthetic data

---

## Role Definition: Embodied Generalist

We're not looking for a pure AI researcher or traditional robotics engineer. We're looking for an Embodied Generalist—someone who can bridge AI, robotics, systems, and deployment to independently get robots running in the real world.

Core Capability Model:

- 40% Robotics - ROS2, Control, Perception, Sensor Fusion

- 30% Systems - Deployment, Optimization, Debugging, Monitoring

- 20% ML/VLA - Model Application, Fine-tuning, Inference Optimization

- 10% Deployment - On-site Delivery, Rapid Iteration

---

## Core Responsibilities: End-to-End Loop

### 1. Data Loop

- Design and implement robot data collection systems

- Process multi-sensor data (cameras, LiDAR, IMU)

- Build robot telemetry and replay mechanisms

- Establish data cleaning and annotation pipelines

### 2. Model Engineering

- Deploy VLA/VLM models to edge devices (NVIDIA Jetson)

- Optimize model inference (quantization, pruning, TensorRT)

- Fine-tune policy models for new scenarios

- Handle Sim2Real migration challenges

### 3. System Integration & Deployment

- Full-stack ROS2 development (messages, services, parameters, behavior trees)

- Multi-sensor calibration and fusion

- Real-time system optimization (latency < 100ms)

- Docker containerized deployment

### 4. On-site Delivery & Debugging

- Rapid deployment in real environments (exhibition halls, factories)

- On-site fault diagnosis and repair

- Customer demo preparation and support

- Operations and monitoring system setup

---

## Required Skills: 40% Robotics (Most Important)

| Skill | Priority | Description |

|------|----------|-------------|

| ROS2 | Extremely High | 3+ years actual project experience, skilled in debugging |

| Linux | Extremely High | System-level debugging and performance optimization |

| C++ | Extremely High | High-quality production code |

| Python | Extremely High | Scripting, data processing, ML toolchain |

| Sensor Integration | Extremely High | Cameras, LiDAR, IMU, multi-sensor fusion |

| Calibration | High | Camera calibration, sensor calibration |

| Motion Control | High | Trajectory planning, kinematics fundamentals |

| Real-time Systems | High | Understanding timing constraints, handling latency |

Bonus Skills:

- Manipulation (robotic arm control)

- SLAM (localization and mapping)

- Navigation (path planning)

- Embedded systems development

- CAN/EtherCAT (industrial buses)

---

## VLA/AI Capabilities: 20% ML/VLA

Not asking you to publish NeurIPS papers, but:

> "Can use + Can modify + Can deploy"

### Must Understand

- Transformer basics - Understanding attention, multimodal fusion

- Multimodal - Vision-Language-Action cross-modal

- Imitation Learning - Behavior cloning, imitation learning

- RL basics - Reinforcement learning fundamentals

- Diffusion Policy - Diffusion models for policy

- VLM/VLA Inference - Inference and optimization

### Expected Capabilities

- ✅ Fine-tune policy models

- ✅ Inference optimization

- ✅ Data cleaning and preprocessing

- ✅ Process robot trajectory data

- ✅ Deploy to edge devices

### Not Required (That's Research Scientist's Job)

- ❌ Create foundation models

- ❌ Publish SOTA papers

---

## Systems Capabilities: 30% Systems

This determines your ceiling in real-world deployment

| Skill | Priority | Description |

|------|----------|-------------|

| Docker | Extremely High | Containerized deployment and environment isolation |

| GPU Profiling | High | Performance analysis and optimization |

| Edge Deployment | Extremely High | Jetson and other edge device deployment |

| Networking | High | Multi-threading, inter-process communication |

| Logging/Telemetry | High | Logging, monitoring, telemetry |

Why This Matters:

Real-world scenarios aren't benchmarks—they're about:

- Latency

- Memory

- Bandwidth

- Inference instability

---

## What We Value

### ✅ Prioritized

- Real robot experience - Papers don't matter if you can't make robots move

- End-to-end capability - Data collection → model training → deployment → closed-loop, full process mastery

- On-site problem solving - Rapidly diagnose and fix issues in factory/exhibition environments

- Cross-layer collaboration - Handle AI, robotics, systems, and deployment simultaneously

- Rapid iteration - Stable delivery under high pressure with changing requirements

### ❌ Not Priority

- Pure paper count (no practical deployment)

- Only training models, no engineering deployment

- Only calling APIs, no understanding of underlying implementation

- Big company prestige without hands-on ability

- Only simulation experience (Isaac Sim/Mujoco), little real robot experience

---

## Why Join Us?

### 1. Real-World Scenarios

Not demo labs, but:

- Actual deployments in Hong Kong malls, residential buildings, factories

- Real robots running 24/7

- Facing noise and uncertainty in the real world

### 2. Rapid Iteration

- Clear customer requirements

- Short feedback cycles

- Your code directly impacts products

### 3. Cutting-Edge Technology

- Latest applications of VLA/VLM

- Multimodal model engineering

- Embodied AI practical deployment

### 4. Core Team

- Early member, directly participate in product definition

- Work closely with Technical Director, rapid growth

- Access to full-stack robotics technology

What We Offer

• Competitive salary and performance‑linked incentives.

• Opportunity to work at the forefront of AI & robotics innovation, interacting with engineers, product teams, and global clients.

• Support for professional development, including training in AI‑driven marketing and emerging technologies.

If you are passionate about technology, enjoy turning complex ideas into compelling marketing messages, and want to grow your career in the AI & Robotics space, we would love to hear from you.

We offer competitive package and promising career prospect to the right candidate. Interested parties please send resume and expected salary by clicking “Apply Now”.

For more information, please visit our website at https://www.axonex.ai/