AI Research Scientist: Multimodal World Models

Datadog; Pittsburgh (Allegheny County); On-site; USD 320,000 - 400,000; Full time

Benefits offered by this job

Stock equity (RSUs)

ESPP

Conference attendance

Mentor program

Inclusive culture

Job summary

Datadog is seeking a Research Scientist in the United States to advance fundamental AI research and translate advances into cloud observability and security products. You will collaborate with Research Engineers and product teams to build world models and autonomous agents operating on multimodal telemetry data.

You will publish at top venues, design scalable training and evaluation frameworks, and help open-source key model artifacts and benchmarks, shaping Datadog's AI-driven product roadmap.

Qualifications

• PhD in Computer Science, ML, or related field.
• Deep learning models and agents with distributed training.
• Publications at top venues (NeurIPS/ICLR/ICML/TMLR).
• Experience with training, post-training, and inference for large foundation models.
• Able to explain complex research to technical and non-technical audiences.

Responsibilities

• Conduct research in generative AI and ML, building foundation models and trained agents for observability.
• Train multimodal models on telemetry data (metrics, logs, traces) using distributed training.
• Design and build simulated environments and RL loops for agent training and evaluation.
• Collaborate with Product/Engineering to integrate capabilities into Datadog products.
• Stay at the forefront of foundation models and RL-based agents.
• Contribute to publications and open-source model artifacts.

Skills

Generative modeling

World models

AI agents

Reinforcement learning

Multimodal learning

Communication to non-technical

Education

PhD in Computer Science, ML, or related field

Tools

DeepSpeed

Megatron-LM

PyTorch

Job description

Datadog is seeking a Research Scientist in the United States to advance fundamental AI research and translate advances into cloud observability and security products. You will collaborate with Research Engineers and product teams to build world models and autonomous agents operating on multimodal telemetry data.

You will publish at top venues, design scalable training and evaluation frameworks, and help open-source key model artifacts and benchmarks, shaping Datadog's AI-driven product roadmap.