Synthetic Visual Data Engineer (Simulation & Domain Randomization) at ANYbotics

Barcelona, Catalonia, Spain

Apply Now

Not SpecifiedCompensation

Mid-level (3 to 4 years), Senior (5 to 8 years)Experience Level

Full TimeJob Type

UnknownVisa

Robotics, Industrial Automation, Energy, UtilitiesIndustries

Requirements

3+ years experience building synthetic data pipelines or strong background in computer graphics + Python tooling
Proficiency in Python and image processing (OpenCV, Pillow); experience with 3D pipelines (Blender scripting, Unity, or equivalent) is highly useful
Experience with generative models (GANs, VAEs, diffusion) or commercial synthetic-data platforms (Synthia, Unity Perception, Rendered.ai)
Practical knowledge of augmentation and domain randomization techniques for vision (lighting, noise, geometric transforms)
Good ML literacy: able to run quick training/validation loops to measure synthetic data impact (PyTorch/TensorFlow)
Strong visual judgment and attention to detail for crafting realistic synthetic examples
Optional requirements
Prior work in privacy-preserving synthetic data, fairness testing, or statistics-based validation of synthetic datasets
Familiarity with asset pipelines (texture/material authoring) and camera/sensor simulation (exposure models, noise)
Experience producing domain-specific synthetic assets (industrial gauges, manufacturing defects, remote-sensing scenes)

Responsibilities

Architect and implement reusable synthetic data generators for each core visual inspection task
Create and curate an asset library (backgrounds, object crops, fonts, CAD/3D assets) and parameterized recipes for domain randomization (lighting, blur, occlusion, material)
Integrate synthetic generation into the dataset pipeline (consistent labels, formats like COCO / segmentation masks / text transcriptions); automate large-batch generation and metadata
Validate synthetic realism and utility by running targeted ML experiments and iterating on generation parameters to reduce sim-to-real gap
Implement privacy- and bias-aware generation practices; document generation configs and ensure reproducibility
Collaborate closely with ML engineers and data engineers to surface data gaps (from telemetry or failure cases) and prioritize synthetic scenarios

Skills

Synthetic Data

Data Pipelines

Domain Randomization

3D Assets

CAD

Computer Vision

Image Annotation

Simulation

LIDAR

Robotics

Python

Asset Library Management

ANYbotics

Develops autonomous robots for industrial inspections

About ANYbotics

ANYbotics specializes in creating autonomous legged robots for industrial inspection. Their main product, ANYmal, is a four-legged robot that can patrol challenging environments to gather and analyze data. This technology is especially useful in industries like oil, gas, and chemicals, where regular inspections are vital for safety and efficiency. ANYbotics serves industrial operators who need to monitor their facilities, offering a solution that is more scalable, safer, and cost-effective than traditional inspection methods. ANYmal can traverse difficult terrains and deliver actionable insights at any time, helping to maximize equipment uptime and minimize operational costs. The company operates in the industrial automation sector, focusing on robotic inspection services, and generates revenue by selling robots and providing ongoing support. The goal of ANYbotics is to enhance safety standards and reduce costs for their clients by automating routine inspection tasks.

Zurich, SwitzerlandHeadquarters

2016Year Founded

$128.7MTotal Funding

LATE_VCCompany Stage

Robotics & Automation, Industrial & ManufacturingIndustries

51-200Employees