AI Engineer (Vision-Language Models)

TL;DR

AI Engineer (Vision-Language Models): Training and fine-tuning multimodal models for video and image understanding with an accent on alignment pipelines, MoE architectures, and production-grade inference optimization. Focus on designing evaluation benchmarks, managing large-scale synthetic data pipelines, and solving complex temporal modeling challenges for AI safety systems.

Location: Must be based in or be able to relocate to Paris (Hybrid)

Salary: $100,000 – $250,000 + equity

Company

hirify.global is an AI safety startup building the reliability and optimization layer for AI systems through natural-language policy enforcement.

What you will do

• Train vision-language models from scratch and fine-tune existing architectures for image understanding.
• Extend VLM capabilities to video by designing temporal modeling approaches and handling long-context data.
• Design and implement evaluation benchmarks for visual QA, spatial reasoning, and 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 with a focus on quantization, batching, and latency optimization.

Requirements

• 3+ years of experience training and fine-tuning vision-language models like LLaVA or Qwen-VL.
• Deep understanding of multimodal architectures, including vision encoders, projectors, and LLMs.
• Hands-on experience with multimodal alignment techniques such as GRPO, DPO, and reward modeling.
• Proven track record of shipping VLM solutions to production environments.
• Strong proficiency in PyTorch and distributed training frameworks like DeepSpeed or FSDP.
• English proficiency: C1 level required for daily operations.

Culture & Benefits

• Comprehensive medical insurance in France.
• Relocation package provided for international candidates.
• Flexible paid time off policy.
• Provision of all necessary hardware, tools, and AI service subscriptions.
• Regular team off-sites held twice a year.

Hiring process

• Introductory call with the team.
• Completion of a take-home assignment.
• Technical interview session.
• Final interview with the CEO and CTO.