ML Engineer (AI)

TL;DR

ML Engineer (AI/LLMs): Training, post-training, and evaluating core LLMs for an AI safety platform with an accent on SFT, RLHF, and DPO-style alignment. Focus on building large-scale data pipelines, distributed multi-GPU training, and optimizing production inference.

Location: Hybrid in Paris or London (Relocation package available for Paris only)

Compensation: $120K – $250K + Equity

Company

AI safety company building a reliability and optimization layer for AI systems using natural-language policies to enforce model behavior.

What you will do

• Train and post-train LLMs using SFT, RLHF, DPO, and related alignment methods.
• Build reward models based on human and synthetic preference data.
• Design and manage high-throughput data pipelines for collection, filtering, and quality control at scale.
• Execute distributed training on multi-GPU clusters and debug performance issues.
• Develop evaluation systems and benchmarks to drive training decisions.
• Optimize models for production inference using quantization, speculative decoding, and vLLM/TensorRT.

Requirements

• Hands-on experience with modern LLM post-training (SFT, RLHF, DPO) on self-trained models.
• Experience building large-scale data pipelines for training corpora and synthetic data.
• Proficiency in PyTorch or JAX with experience in distributed multi-GPU training.
• Deep understanding of model evaluation and the ability to build reliable benchmarks.
• Experience with inference optimization tools like vLLM, TensorRT, or Triton.
• Must be based in or able to relocate to Paris or London.

Nice to have

• Public builder footprint (open-source models, datasets, or papers on HuggingFace/GitHub).
• Experience at frontier or near-frontier AI labs.
• Knowledge of advanced RL methods for LLMs (e.g., online RL, GRPO-style methods).
• Experience with large-scale moderation, safety, or classification models.
• Experience in multilingual model training.

Culture & Benefits

• Paid time off in accordance with local regulations.
• Comprehensive medical insurance for the France-based team.
• Relocation support available for candidates moving to Paris.
• Full provision of necessary hardware, AI agent subscriptions, and IDEs.
• Bi-annual team off-sites in diverse locations.

Hiring process

• Introductory call with HR (25 min).
• Take-home technical test task.
• Technical interview with the Head of Applied Research (60 min).
• Final conversation with the CEO (45 min).