Multimodal ML Engineer

Responsibilities

We are representing a confidential AI safety organization seeking a Multimodal ML Engineer to join their Paris-based team. In this role, you will:

• Train and fine-tune large-scale multimodal models including vision-language, audio, and speech systems from scratch and from pretrained checkpoints
• Extend models across modalities: image understanding, video temporal modeling, long-context processing, and streaming audio
• Design and execute experiments around architecture changes, data composition, and training recipes
• Build and maintain multimodal data pipelines from raw images, video, and audio recordings to training-ready datasets, including synthetic data generation
• Train and optimize Mixture of Experts (MoE) architectures for efficient multimodal inference
• Develop alignment pipelines including supervised fine-tuning, direct preference optimization, and reward modeling across multiple modalities
• Optimize models for production through quantization, distillation, batching, and low-latency serving
• Deploy models end-to-end from research checkpoint to production systems
• Define evaluation metrics and benchmarks for visual QA, spatial reasoning, video comprehension, and speech understanding

What We're Looking For

Our client needs someone with:

• 3+ years of experience training large-scale deep learning models in multimodal domains (vision-language, audio, speech, or acoustic)
• Strong PyTorch skills with hands-on distributed training experience using DeepSpeed, FSDP, or similar frameworks
• Deep experience with multimodal architectures and understanding of how vision/audio encoders, projectors, and language models integrate
• Hands-on experience with RLHF and alignment techniques for multimodal systems, including GRPO, DPO, and reward modeling beyond text-only applications
• Experience with video and/or audio sequence modeling: temporal modeling, long-context processing, efficient attention mechanisms, and streaming inference
• Proven track record of shipping models to production, meeting latency targets, and optimizing inference performance
• Comfort with large-scale multimodal dataset curation including image-text pairs, video-instruction data, audio preprocessing, augmentation, and synthetic data generation
• Familiarity with MoE architectures and their tradeoffs for multimodal workloads
• Strong engineering fundamentals: clean code, version control, testing, and documentation

Bonus qualifications: Understanding of audio signal processing fundamentals including spectrograms, mel features, and noise reduction.

How We Work With You

Our talent team at Get A Job.ai will guide you through this confidential search:

• Apply directly through the Get A Job.ai platform with your resume and relevant portfolio or GitHub links
• A recruiter from our team will conduct an initial screening to understand your background and motivations
• We will submit qualified candidates to our client for consideration
• Please do not attempt to contact the client directly, as this is a confidential search
• Our client's interview process includes a take-home technical task, a technical interview with their Head of Applied Research, and a final conversation with leadership

Location and Benefits

This position is based in Paris with hybrid work arrangements. A relocation package is available for qualified candidates. The client offers:

• Competitive paid time off in line with local regulations
• Comprehensive medical insurance for France-based team members
• Meaningful equity package
• All necessary hardware, tools, and services
• Covered subscriptions for AI development tools
• Team off-site events twice annually

Pay

Compensation details will be discussed during the screening process with Get A Job.ai and are competitive for senior multimodal ML engineering roles in the Paris market.

Equal Opportunity: Get A Job.ai is committed to inclusive recruiting practices. We partner with clients who value diversity and welcome applicants from all backgrounds.