Member of Technical Staff (Foundation Models)

Role overview

Work across research and engineering to build foundation models and AI systems that learn how the world changes over time. The role focuses on temporal intelligence: combining observations, forecasts, uncertainty, causal reasoning, and decision support so models can move beyond describing the past and help determine what may happen next.

Responsibilities

- Design and train large-scale temporal foundation models using heterogeneous time-series and structured datasets, with a stated target of scaling toward more than 10 trillion observations. - Develop multimodal models that combine text, time series, tabular data, events, and other structured signals for forecasting and interpretation. - Build agentic forecasting systems that break down questions, retrieve relevant information, use models and tools, reason about uncertainty, and synthesise answers. - Develop efficient training and inference pipelines for low-latency, high-throughput forecasting across many models, datasets, horizons, and users. - Advance methods connecting forecasting with causal learning, counterfactual reasoning, uncertainty estimation, and intervention analysis. - Adapt general temporal models through pre-training, post-training, fine-tuning, and other techniques, then translate research into working systems across the ML stack.

Requirements

- Strong research intuition combined with the ability to build and ship functioning ML systems. - Ability to contribute across data, model architecture, distributed training, evaluation, inference, and deployment. - Interest or experience in understanding temporal dynamics, forecasting, uncertainty, and decisions under changing conditions.

Nice to have

- Experience training foundation models from scratch or working with distributed training, large-scale data pipelines, GPU clusters, and model optimisation. - Research experience in time-series foundation models, forecasting, multimodal learning, causal learning, or generative modelling. - Experience with efficient inference, model serving, or agents combining language models, forecasting models, retrieval, code execution, and external tools. - Publications at major machine-learning venues or significant contributions to widely used open-source ML systems.

Benefits and work setup

This is a full-time research and engineering role offered remotely across the United States and globally.