Hybrid AI Engineer, Vision

at White Circle

• Job title: AI Engineer, Vision
• Job type: Full Time
• Experience level: Mid level, Senior
• Salary: $100,000 - $250,000 per year
• Degree requirement: No Education Requirement
• Tech skills: PyTorch
• Location requirements: Hybrid, Paris, France

About the role

• AI Engineer responsible for training vision-language models for AI Safety company in Paris. Extending capabilities to video and optimizing inference for production environments.

Responsibilities

• Train vision-language models from scratch and fine-tune existing architectures for image understanding
• Extend VLM capabilities to video: design temporal modeling approaches, handle long-context efficiently
• Design evaluation benchmarks that matter: visual QA, spatial reasoning, video comprehension
• Curate and maintain multimodal datasets — including synthetic data generation pipelines
• Train and optimize MoE architectures for efficient multimodal inference
• Deploy models to production: quantization, batching strategies, latency optimization

Requirements

• 3+ years training and fine-tuning vision-language models (LLaVA, Qwen-VL, InternVL, or similar)
• Deep experience with multimodal architectures — you understand how vision encoders, projectors, and LLMs fit together
• Hands-on with RLHF/alignment for multimodal: GRPO, DPO, reward modeling — not just for text
• Experience with video understanding: temporal modeling, long-context processing, efficient attention mechanisms
• Track record shipping VLMs to production: you've optimized inference, not just reported benchmark scores
• Comfortable with large-scale dataset curation: image-text pairs, video-instruction data, synthetic data generation
• Familiar with MoE architectures and their tradeoffs for multimodal workloads
• Strong PyTorch skills, experience with distributed training (DeepSpeed, FSDP)

Benefits

• 20 days of paid vacation
• Work from Paris (hybrid) + relocation package
• Best medical insurance in France
• All the hardware, tools, and services you need
• Covered subscriptions for AI agents and IDEs
• Team off-sites twice a year: we’ve recently been to the Alps and to Saint-Tropez