AI Research Scientist - Datadog AI Research (DAIR)

The role

This role sits within Datadog's AI research group, focused on high-risk, high-payoff work tied to observability and security problems. The scientist partners with research engineers to push fundamental ideas and then carry them into Datadog's products alongside product and engineering colleagues.

What you would do

• Run generative AI and machine learning research aimed at specialized models and autonomous agents for monitoring systems
• Train multimodal models over large, heterogeneous telemetry streams using distributed compute
• Create simulated settings and reinforcement learning loops for training and benchmarking agents
• Work with product and engineering groups to embed world-modeling and agent capabilities into shipped features
• Track developments in foundation models, world models, and RL-driven agents
• Publish findings at leading conferences and help release models and benchmarks as open source

What they asked for

• Doctoral degree in computer science, machine learning, or an adjacent field, or comparable practical experience
• Deep background in generative modeling, world models, agents, reinforcement learning, or multimodal learning
• Substantial hands-on work building deep learning models and agents with distributed training stacks and PyTorch
• Record of impactful papers at top venues such as NeurIPS, ICLR, ICML, or TMLR
• Working knowledge of efficient training, post-training, and inference methods for large foundation models
• Ability to communicate complex research clearly to both technical and general audiences

Nice to have

• Experience moving research into real product use, particularly with foundation models, world models, or RL agents
• Drive to advance AI with attention to customer value and scalable rollout
• Experience building production data pipelines and applications
• Hands-on GPU programming and optimization work, including CUDA