Robotics Software Engineer – Machine Learning, Computer Vision & VLM

United Robots is a technology company developing autonomous mobile robots for industrial facilities, warehouses, and logistics centres. We create complete solutions encompassing electronics, control systems, autonomous navigation, sensor integration, robot software, and a cloud platform for managing fleets of devices.

We are looking for a Robotics Software Engineer specialising in machine learning and computer vision. In this role, you will develop algorithms that enable robots to recognise their surroundings, people, vehicles, obstacles, and events relevant to safe and efficient operation.

Responsibilities

• Developing computer-vision and machine-learning algorithms for autonomous robots;

• Designing and implementing object detection, classification, segmentation, and tracking systems;

• Developing algorithms for detecting people, forklifts, vehicles, and obstacles;

• Developing semantic understanding of the robot’s environment;

• Processing data from RGB, RGB-D, stereo, thermal, and other cameras;

• Combining visual information with LiDAR and other sensor data;

• Preparing, cleaning, annotating, and versioning datasets;

• Designing model-training, validation, and comparison processes;

• Selecting model architectures and learning methods appropriate to application constraints;

• Optimising models for real-time operation on onboard computers;

• Converting and deploying models on edge-computing platforms;

• Developing mechanisms for monitoring model quality on deployed robots;

• Detecting model degradation and unusual cases;

• Creating tools for automated analysis of data collected by the robot fleet;

• Developing perception components in ROS and ROS 2;

• Integrating perception outputs with navigation, safety, and mission-execution systems;

• Testing algorithms in simulation and on physical robots;

• Analysing issues arising under changing lighting and environmental conditions;

• Creating automated tests and model-quality metrics;

• Preparing technical documentation;

• Collaborating with SLAM, navigation, embedded, and cloud-platform teams.

Key objectives of the role

• Increasing robots’ ability to understand dynamic environments;

• Improving the detection of people, vehicles, and obstacles;

• Ensuring stable algorithm performance under changing lighting conditions;

• Reducing false detections and missed objects;

• Implementing a repeatable model-training, testing, and publishing process;

• Optimising models for real-time operation on edge devices;

• Using fleet data to continuously improve the system;

• Developing shared perception components for different robot models.

Requirements

• A minimum of 2 years of relevant professional experience;

• Experience in machine learning, deep learning, or computer vision projects;

• Very good knowledge of Python;

• Knowledge of at least one framework: PyTorch, TensorFlow, or JAX;

• Practical knowledge of OpenCV;

• Experience with object detection, classification, segmentation, or tracking;

• Knowledge of dataset preparation and validation methods;

• Ability to define appropriate model-quality metrics;

• Knowledge of generalisation, overfitting, and data augmentation;

• Experience deploying models to production;

• Knowledge of Linux and Git;

• Ability to analyse model errors and edge cases;

• Ability to work with data from real sensors;

• English proficiency sufficient to work confidently with documentation and research papers.

Nice to have

• Knowledge of C++;

• Experience with ROS or ROS 2;

• Experience with NVIDIA Jetson, CUDA, TensorRT, or DeepStream;

• Knowledge of ONNX and model-optimisation tools;

• Experience with YOLO, Detectron2, MMDetection, or similar solutions;

• Knowledge of Vision Transformers and modern multimodal models, including Vision-Language Models (VLMs), prompt design, fine-tuning, and evaluation;

• Experience with RGB-D, stereo, or thermal cameras;

• Knowledge of sensor fusion and projection between coordinate frames;

• Experience processing point clouds;

• Knowledge of MLOps, data and model versioning, and experiment tracking;

• Experience with Docker, CI/CD, and automated model testing;

• Knowledge of synthetic data and simulation environments;

• Experience with active learning, continual learning, or domain adaptation;

• Experience in mobile robotics, automotive, or safety systems;

• Publications, open-source projects, or a computer-vision portfolio.

What we offer

• A genuine influence on the perception system of autonomous robots;

• Work with real data collected by robots operating at customer sites;

• The opportunity to deploy models on physical edge platforms;

• Access to cameras, LiDARs, robots, and a dedicated testing environment;

• The opportunity to build a complete model-development process from data to fleet deployment;

• Work combining robotics, computer vision, machine learning, and cloud systems;

• A high degree of independence and direct cooperation with the technical team;

• Compensation aligned with your experience and level of responsibility.