AI Research Scientist

About the role

The Datadog AI Research (DAIR) team is seeking a scientist to advance generative AI and machine learning solutions for cloud observability and security. You will work alongside research engineers to develop foundation models and autonomous agents that solve complex real-world infrastructure challenges.

Key facts

• Location: Paris, France
• Engagement: Full-time
• Team: Datadog AI Research (DAIR)

What you'll do

• Conduct research in generative AI and machine learning to build specialized foundation models and agents.
• Train multimodal models on diverse telemetry data including logs, traces, metrics, topology, and events using distributed infrastructure.
• Develop simulated environments and reinforcement learning training loops for on-policy agent evaluation and training.
• Partner with product and engineering departments to integrate world modeling and agent capabilities into Datadog products.
• Publish research findings, present at major conferences such as NeurIPS, ICLR, and ICML, and contribute to open-source benchmarks.
• Communicate complex research and model mechanics to both technical and non-technical stakeholders.

Requirements

• PhD in Computer Science, Machine Learning, or a related field, or equivalent experience.
• Expertise in reinforcement learning, world models, generative modeling, AI agents, or multimodal learning.
• Proficiency in PyTorch and distributed training frameworks like Megatron-LM or DeepSpeed.
• Proven track record of impactful research publications at top-tier venues like NeurIPS, ICLR, ICML, or TMLR.
• Knowledge of inference, post-training, and efficient training techniques for large foundation models.

Nice to have

• Experience bridging research with real-world product applications.
• Background in writing production-grade data pipelines and applications.
• Hands-on experience with GPU optimization and CUDA programming.

Skills & tools

PyTorch - DeepSpeed - Megatron-LM - Distributed training frameworks - Reinforcement Learning - Generative AI - Multimodal learning

Practical notes

• Compensation includes competitive global benefits, new hire stock equity (RSUs), and an employee stock purchase plan (ESPP).
• Role involves collaboration with teams in Paris and New York City.
• Datadog provides opportunities for conference attendance and participation in internal mentor programs.
• Requisition ID: R12618.