Jobs at Ai2

Develop an open-source AI-powered, physics-informed climate model for improved multidecadal forecasting. Adapt NeuralGCM training pipelines for Google Cloud and NVIDIA GPUs, document and distribute code, couple the model with an ML ocean model, test performance in unseen climates, collaborate through software engineering practices, and create tutorials for the climate modeling community.

Lead research and evaluation of an AI-powered, physics-informed climate model designed for accurate multidecadal forecasts. Develop training and testing cases, implement model improvements for generalization to unseen climates, couple atmospheric, oceanic, land, and sea-ice systems, publish findings, contribute open-source code, collaborate with researchers, and mentor interns.

Develop an AI-powered, physics-informed hybrid climate model for accurate multidecadal forecasting. Responsibilities include coupling NeuralGCM with an ML ocean model, designing physically constrained learning approaches, testing model performance in unseen climates, publishing research, contributing to open-source code, collaborating with climate scientists and engineers, and mentoring postdocs and interns.

Own product strategy and execution for Asta, an agentic scientific research platform, including engineering frameworks, data infrastructure, public APIs, benchmarks, and compliance. Translate researcher needs into requirements and roadmaps, drive adoption, coordinate engineering workstreams, and maintain the Semantic Scholar platform for millions of users. Partner across research, engineering, legal, and leadership while shaping product management practices.

Lead research and scientific development for fully open foundation models focused on scientific discovery. Build and train modern foundation models, develop training pipelines, curate high-quality data, and investigate agentic and multimodal training. Provide scientific leadership, lead small teams, oversee research software systems, publish and present peer-reviewed work, and build collaborations with academic, industry, government, and standards organizations.

Lead product design for AI-powered environmental conservation and Earth-observation products. Conduct user research, solve complex data-heavy interaction problems, and design interfaces involving maps, dashboards, model training, evaluation metrics, agents, and geospatial data. Own work end to end from problem framing through shipping and iteration, while building design systems and collaborating closely with product, engineering, ML research, and global users.

Lead end-to-end development and deployment of large open models (Olmo, Molmo), building and optimizing training/inference pipelines, dataset curation, and large-scale preprocessing; design, train, and evaluate multimodal and agentic systems; scope and run high-impact research projects; apply strong software engineering to bridge research to production and contribute to open-source releases.

The Research Scientist will develop core research problems in robotics, engage in simulations, lead projects, mentor interns, and collaborate widely.

The role involves developing and maintaining robotic control stacks, managing sim-to-real pipelines, and collaborating with researchers on robotic learning projects.

The Research Scientist will develop AI systems with deep reasoning capabilities, contribute to open-source research, and mentor early-career researchers. Responsibilities include leading foundational research, authoring technical reports, collaborating with teams, and advancing agentic systems.

Collaborate with Ai2 researchers on language-model and NLP research, gain hands-on experience with Ai2's data and engineering infrastructure, co-author papers for top venues, and prepare for PhD studies. Work is research-focused with mentorship and no teaching/admin duties. On-site work in Seattle is expected.