Siddharth Narayanan

Research

Narayanan is first author of Aviary (December 2024), a FutureHouse framework for training language agents on challenging scientific tasks, and contributes to its open-source repositories.⁴⁵⁶⁷

In June 2025, Narayanan was first author of "Training a Scientific Reasoning Model for Chemistry", which lists FutureHouse as the authors' affiliation and introduced ether0, a 24B-parameter reasoning model for chemistry released with open weights. Narayanan's GitHub profile now lists Edison Scientific.²³⁸¹

Work and contributions

• ether0 (24B) Open weights

  First author of the paper introducing ether0, a 24B-parameter model trained with reinforcement learning to reason in natural language and answer with chemical structures.²³⁸

• Aviary Open-source code

  First author of the Aviary paper (2024) and a contributor to the open-source Aviary and LDP repositories for building and training language agents.⁴⁵⁶⁷

Timeline

1. Dec 2024Posted the Aviary paper on training language agents for scientific tasks as first author.⁴
2. Jun 2025Posted "Training a Scientific Reasoning Model for Chemistry", introducing ether0, as first author.²

Other documented roles

Roles the sources document that are not confirmed as current.

• Researcher (affiliation on the 2025 ether0 paper), FutureHouse³

Where they work

• Edison Scientific 3 people in the wiki

Related people

• Sam RodriquesAlso at Edison Scientific
• Andrew WhiteAlso at Edison Scientific
• Niko GrupenAgents · Reinforcement learning
• Shimon WhitesonReinforcement learning · Agents
• Jeff CluneReinforcement learning · Agents
• Josh AlbrechtAgents · Reinforcement learning
• Kenneth StanleyReinforcement learning · Agents
• James BrazaAI for science · Agents
• David HaReinforcement learning · Agents
• Pushmeet KohliAI for science · Agents