ML Research Engineer

United States Digital Space LLC; Paris (TX); Hybrid; USD 103,000 - 137,000; Full time

Benefits offered by this job

Hybrid work in Paris/London

Relocation support for Paris

Private health insurance

L&D budget

Team off-sites

Job summary

White Circle seeks ML Engineers to join its AI Safety team building policy enforcement and optimization layers for AI systems. Relocation to Paris or London (hybrid) is required, with hybrid work options and relocation support.

Strong Python and SQL skills, production-grade NLP/ML on large-scale text, and experience with scalable data pipelines are essential. Prior safety/moderation datasets experience is a plus.

Qualifications

• Production-grade pipelines in Python/SQL.
• Applied NLP/ML on real-world text.
• Experience at scale with distributed processing and data storage.

Responsibilities

• Turn petabytes of unstructured text into structured views: topics, clusters, segments, trends, anomalies.
• Build scalable representation pipelines: sampling, preprocessing, embeddings, indexing, retrieval.
• Use LLMs for labeling, weak supervision, data enrichment, and diagnostics with cost/quality controls.
• Translate findings into product and operational decisions; ship self-serve datasets and dashboards.
• Collaborate with engineering and research to align pipelines with production constraints (latency, cost, privacy).

Skills

Python

SQL

NLP

Distributed processing

Model evaluation

Job description

We're looking for ML Engineers to join White Circle, an AI Safety company building the policy enforcement and optimization layer for AI systems. Backed by $11M from senior leaders at OpenAI, Anthropic, HuggingFace, Mistral, and DeepMind, White Circle processes 100M+ API calls monthly and runs its own LLMs in production.

You will

• Turn petabytes of unstructured text into a structured, explorable view: topics, clusters, segments, trends, anomalies.
• Build scalable representation pipelines: sampling, preprocessing, embeddings at scale, indexing, and retrieval.
• Use LLMs for labeling, weak supervision, data enrichment, and automated diagnostics, with cost/quality controls.
• Translate findings into product and operational decisions, and ship self-serve datasets, data models, and dashboards.
• Work with engineering and research to align pipelines with production constraints (latency, cost, privacy).

Requirements

• Strong Python and SQL, with production-grade pipeline engineering (not just notebooks).
• Applied NLP/ML on real-world text: embeddings, clustering, topic modeling, semantic search, classification.
• Experience at scale: distributed processing, large-scale storage and querying, performance-cost tradeoffs.
• Evaluation of fuzzy problems: offline/online metrics, human-in-the-loop labeling, inter-annotator agreement, drift monitoring.
• Prior work with safety/moderation datasets, policy/rule systems, or high-volume logging/observability.
• Relocation to Paris or London (hybrid) required.

Bonus

• Public builder footprint: open-source models, datasets, or frameworks on HuggingFace/GitHub, papers, or technical posts.
• Experience at a frontier or near-frontier lab, or leading open-source model releases.
• RL for LLMs beyond standard RLHF: online RL, GRPO-style methods.
• Moderation, safety, or classification models at scale; multilingual model training.

We offer

• Competitive salary + equity.
• Hybrid work from central London or Paris office, relocation support for Paris after probation.
• Premium private health insurance, mental health support, flexible time off.
• Lunch and dinner covered in the office, L&D budget, all hardware and tools you need.
• Team off-sites twice a year.