Research Engineer: AI Agents & Open-Source Models

Dayhoff Labs; Cambridge (MA); On-site; USD 150,000 - 210,000; Full time

Job summary

Dayhoff Labs is seeking a generalist research engineer to span maintenance and research, using LLMs and agents to build software and tackle open-ended scientific questions. You’ll have latitude to decide what to build and how your work accelerates researchers’ progress.

You’ll self-host and evaluate open-source models, write the harnesses and tooling to make agents reliable, and contribute across internal tooling, data, and model work as priorities shift.

Qualifications

• Demonstrated ability to use LLMs to write software and build complex systems.
• Hands-on experience with open-source models: self-hosting, evaluation, and writing harnesses or scaffolding around them.
• Strong Python and the ability to own complex technical systems end to end.
• Comfort working across an unfamiliar codebase and shipping in areas outside your prior expertise.
• Sound engineering fundamentals: version control workflows, testing, and reproducible environments.

Responsibilities

• Use LLMs and coding agents to build software and complex systems, and keep advancing the internal agentic systems we use to write our code and maintain our infrastructure
• Self-host and evaluate open-source models, and write the harnesses, tooling, and scaffolding that make agents reliable for our workflows
• Apply agents to scientific questions - an open-ended remit spanning genuine research, experimentation, and analysis, not only software development
• Contribute across a variety of maintenance and research projects as priorities shift, from internal tooling to data and model work
• Keep our cloud infrastructure and development environments healthy

Skills

LLM-driven software

Open-source models

Python

End-to-end systems

Engineering fundamentals

Tools

Self-hosting

Harnesses & scaffolding

Cloud infrastructure

Job description

Dayhoff Labs is seeking a generalist research engineer to span maintenance and research, using LLMs and agents to build software and tackle open-ended scientific questions. You’ll have latitude to decide what to build and how your work accelerates researchers’ progress.

You’ll self-host and evaluate open-source models, write the harnesses and tooling to make agents reliable, and contribute across internal tooling, data, and model work as priorities shift.