Research Scientist / AI Engineer Intern — Temporal Intelligence

Role overview

Own a focused research or engineering project at the intersection of foundation models, time-series analysis, multimodal AI, causal reasoning, and agentic systems. Working closely with a research and engineering team, the intern will contribute to core models or infrastructure and investigate how AI can understand temporal dynamics, forecast future outcomes, and support decisions under uncertainty.

Responsibilities

- Train and evaluate temporal foundation models on datasets scaling toward 10 trillion or more observations, studying the effects of model, data, and compute scaling. - Develop multimodal models that combine text, time series, tabular data, events, and other structured signals. - Build forecasting agents that retrieve information, use models and tools, reason about uncertainty, and produce synthesised answers. - Explore pre-training, post-training, fine-tuning, and adaptation methods for temporal and multimodal foundation models. - Investigate causal discovery, causal inference, and counterfactual reasoning as tools for better forecasting and decision-making. - Design experiments and benchmarks to identify where temporal models succeed or fail and improve their reliability and generalisation.

Requirements

- Ability to understand difficult technical problems, build working systems, run rigorous experiments, and learn quickly. - Practical interest or experience in forecasting, temporal modelling, foundation models, or related AI research and engineering. - Willingness to take ownership of a defined project while collaborating closely with researchers and engineers.

Nice to have

- Experience training or fine-tuning language models, time-series foundation models, or other large-scale models. - Experience with large time-series or tabular datasets, distributed training, GPU clusters, efficient inference, or ML systems optimisation. - Hands-on work with AI agents involving tool use, retrieval, planning, code execution, or model orchestration. - Research or open-source contributions in forecasting, temporal modelling, causal learning, multimodal learning, or generative modelling. Publications are useful but not required.

Benefits and work setup

This is a remote internship available across the United States and globally. The project is connected to core models and infrastructure, with potential contributions to research, open-source systems, and production models rather than an isolated exercise.