Synthetic Data Engineer: Diffusion Models for Vision

NestAI; Helsinki; On-site; EUR 120,000 - 160,000; Full time

Benefits offered by this job

Occupational healthcare

Epassi Flex benefit

Lunch benefit

Team lunches

Mac or PC choice

Holiday allowance

Phone benefit

Job summary

NestAI is building an open, modular platform for sovereign European defense AI. Join the synthetic data team to develop pipelines that generate photorealistic data for robust object detection in defence contexts.

You will train diffusion-based models, including LoRA fine-tuning, and evaluate synthetic data by its impact on real-model performance. This role focuses on research-driven data generation to close data gaps.

Qualifications

• Several years of hands-on experience training generative models yourself.
• Experience creating synthetic data in domains with limited real data.
• Deep familiarity with diffusion models and Python ML training frameworks, including LoRA and Hugging Face Diffusers.

Responsibilities

• Build and improve generative pipelines that produce synthetic image and video data for object detection models.
• Train and fine-tune diffusion-based generative models, including LoRA fine-tuning.
• Combine generative approaches with other data sources to boost realism and variation.
• Validate synthetic data against real-world outcomes to prove training benefits.
• Track new open-source models and evaluate landscape shifts.
• Collaborate with ML Platform chapter and simulation team on data infrastructure.

Skills

Hands-on generative modeling

Synthetic data creation

Diffusion models

LoRA training

Python ML frameworks

Judgment on data realism

Umgang with imperfect information

Tools

Hugging Face Diffusers

LoRA training

Job description

NestAI is building an open, modular platform for sovereign European defense AI. Join the synthetic data team to develop pipelines that generate photorealistic data for robust object detection in defence contexts.

You will train diffusion-based models, including LoRA fine-tuning, and evaluate synthetic data by its impact on real-model performance. This role focuses on research-driven data generation to close data gaps.