Computer Vision AI Scientist - Synthetic Data Generation (m/f/d)

Resaro AI; Deutschland; Vor Ort; EUR 65.000 - 95.000; Vollzeit

Zusammenfassung

A pioneering global AI Assurance company is seeking a Computer Vision AI Scientist to drive synthetic data generation efforts across various formats. This role involves tracking cutting-edge techniques, implementing solutions as modular services, and ensuring continuous improvement in a collaborative environment. Ideal candidates will have a strong foundation in computer vision and machine learning, with a focus on synthetic data techniques. Join a team dedicated to advancing AI technology responsibly and ethically.

Qualifikationen

• Hands-on experience with synthetic data models, including deepfakes, GANs, and diffusion.
• Experience curating datasets and fine-tuning models.
• Familiarity with ComfyUI or similar ML workflow systems.

Aufgaben

• Lead research monitoring and prioritize advancements in synthetic data generation.
• Implement and package models into Docker containers for use.
• Collaborate with ML engineers and report findings.

Kenntnisse

Computer Vision

Machine Learning

Synthetic Data Generation

Python

PyTorch

TensorFlow

Docker

Ausbildung

MSc or PhD in related field

Jobbeschreibung

Resaro was founded on the belief that AI will change the world in ways we cannot even imagine, but every new technology needs safeguards to advance.

About the Job

We're looking for a Computer Vision AI Scientist - Synthetic Data Generation who will lead our efforts in synthetic data generation for images, video, and audio. Your mission is to track and prioritize state-of-the-art synthetic data techniques, implement and fine-tune them, and deploy them as modular Dockerised services integrated with a ComfyUI-based workflow.

The types of data generation include (not exhaustive):

• Deepfake generation (face swaps, voice cloning, lip-syncing, text to image / video).
• Synthetic X-ray scanner imagery.
• Object insertion & spatial orientation control.
• Visual robustness artefacts (e.g., lighting changes, occlusions, adversarial patterns).

You'll manage the full lifecycle of these models from research tracking and model acquisition to fine-tuning, deployment, and ongoing maintenance.

Key Responsibilities

• Research Monitoring & Prioritization
  
  • Track advancements in synthetic data generation (GANs, diffusion models, multimodal deepfakes, etc.).
  • Prioritize which techniques to prototype and productionize based on internal needs.
• Implementation & Packaging
  • Download, set up, and test open-source and pre-trained models.
  • Curate or generate training datasets aligned with test requirements.
  • Fine-tune models on specific tasks or constraints (e.g., realistic object placement, X-ray realism).
  • Optimize models for inference (e.g., ONNX, quantization, batching strategies).
  • Package the solution into Docker containers callable via ComfyUI workflows.
  • Document usage guidelines, performance benchmarks, and known limitations.
• Maintenance & Continuous Improvement
  • Periodically review and update existing models to reflect advances in the field.
  • Ensure compatibility with evolving infrastructure and engineering standards.
• Collaboration & Reporting
  • Work closely with ML engineers, test designers, and UI developers.
  • Share insights with the broader team on model capabilities, trade-offs, and performance.

Required Skills and Experience

• Solid background in computer vision and machine learning, ideally MSc or PhD level.
• Hands-on experience with synthetic data generation models, including deepfakes (image, audio, video), GANs, diffusion, etc.
• Proficiency in Python, PyTorch or TensorFlow.
• Experience curating datasets and fine-tuning models on domain-specific data.
• Comfortable with Docker and building production-ready containers.
• Familiarity with ComfyUI or similar node-based ML workflow systems.
• Experience documenting model performance and integrating models into broader pipelines.
• Ability to independently stay up to date with fast-moving research.

Nice-to-Have

• Experience with audio synthesis and lip-sync tools (e.g., wav2lip, voice cloning).
• Exposure to medical imaging, X-ray simulation, or other imagery types such as thermal/infra-red or LIDAR.
• Familiarity with 3D data manipulation, object insertion in spatially aware contexts
• Knowledge of robustness testing or adversarial ML techniques.
• Experience contributing to open-source projects or publishing research.
• Familiarity with model optimization frameworks (e.g., ONNX, TensorRT).

ABOUT US

Resaro is a global AI Assurance company, pioneering the field of AI testing and evaluation. We are a team of AI experts, engineers, and data scientists. Our mission is to ensure an AI market worthy of trust.

Resaro is an Equal Opportunity Employer. We respect each individual and support the diverse cultures, perspectives, skills and experiences within our teams.