ML Research Engineer
Salary not disclosedHybrid
- Engineering
- Paris
- Full time
- 4d ago
About the role
White Circle is an AI safety company building a reliability and optimization layer for AI systems through natural-language policies. The company processes over 100 million API calls monthly and trains its own LLMs to improve performance and cost-efficiency. As an ML Research Engineer, you will build scalable data pipelines, develop representation models, and use LLMs for automated diagnostics and data enrichment.
Responsibilities
- Turn petabytes of unstructured text into structured views including topics, clusters, and trends.
- Build scalable representation pipelines covering sampling, preprocessing, embeddings, indexing, and retrieval.
- Use LLMs for labeling, weak supervision, data enrichment, and automated diagnostics with cost and quality controls.
- Translate findings into product and operational decisions while shipping self-serve datasets and dashboards.
- Collaborate with engineering and research teams to align pipelines with production constraints such as latency, cost, and privacy.
Required skills
- Python
- SQL
- NLP
- Machine Learning
- Embeddings
- Clustering
- Topic modeling
- Semantic search
- Classification
- Distributed processing
- Large-scale storage
- Drift monitoring
Nice to have
- Open-source model development
- RLHF
- Online RL
- GRPO
- Multilingual model training
Qualifications
- Applied NLP/ML experience on real-world text
- Experience with production-grade pipeline engineering
- Experience with large-scale distributed systems
- Experience with safety or moderation datasets
Benefits
- Equity
- Private health insurance
- Mental health support
- Flexible time off
- Lunch and dinner provided
- Learning and development budget
- Relocation support
- Team off-sites
About the Company
White Circle is an AI safety company backed by 70 million dollars in Series A funding from leaders at OpenAI, Anthropic, HuggingFace, Mistral, and DeepMind. The company builds the safety, reliability, and optimization layer for AI systems through natural-language policies.