Multimodal ML Engineer
Salary not disclosed3+ yearsOnsite
- Engineering
- Paris
- Full time
- 4d ago
About the role
The Multimodal ML Engineer will join an AI safety company to build the reliability and optimization layer for AI systems. You will be responsible for training and fine-tuning large-scale multimodal models, designing experiments, and building data pipelines. The role involves optimizing models for production and defining critical evaluation metrics.
Responsibilities
- Train and fine-tune large-scale multimodal models including vision-language, audio, speech, and video.
- Design experiments, build multimodal data pipelines, and train MoE architectures.
- Build alignment pipelines such as SFT, DPO, and GRPO.
- Optimize models for production through quantization, distillation, and streaming.
- Define evaluation metrics for product performance.
Required skills
- PyTorch
- Distributed training
- DeepSpeed
- FSDP
- LLaVA
- Qwen-VL
- InternVL
- Audio Flamingo
- Whisper
- HuBERT
- Conformer
- RLHF
- GRPO
- DPO
- Sequence modeling
Nice to have
- Audio signal processing
- Spectrograms
- Mel features
- Noise reduction
- MoE architecture
Benefits
- Competitive salary
- Equity
- Visa support
- Relocation assistance
About the Company
White Circle is an AI safety company building the safety, reliability, and optimization layer for AI systems through natural-language policies. The company is backed by $70M in Series A funding and processes over 100M API calls monthly.