Lightcone Research is a team of scientists and engineers on a mission to build the infrastructure for a new era of research, one in which agentic AI is harnessed in the service of rigorous, reproducible, and open science. With a presence at Berkeley and at CNRS in Paris, our team is embedded within some of the world's most prestigious research institutions. Come build the future of AI for science with us!
About the role
Research Software Engineers (RSEs) bring software engineering expertise and open-source development practices to Lightcone's developer and user community. They play the role of an "expert collaborator" who solves problems at the intersection of research, software, and modern computing infrastructure: investigating and proposing solutions to open-ended, unstructured challenges arising from the computational needs of researchers, and scoping and delivering on technical requirements that call for in-depth evaluation of cutting-edge data science and AI applications. RSEs are integral members of a diverse team, co-creating the Lightcone ecosystem of open-source tools to solve real-world challenges in AI-assisted scientific research.
This role works closely with senior researchers and industry partners, advocates for open science and open scholarship best practices, and showcases research outputs and software tools through presentations, research papers, blog posts, interactive data visualizations, and open-source software packages.
What you'll do
- Plan, design, develop, debug, deploy, and evaluate highly complex interface, software, and infrastructure solutions that enable agentic AI systems to produce reproducible scientific research (40%)
- Contribute to and sustain Lightcone's open-source projects: issue triage, code and documentation review, new features, mentoring contributors, CI testing, release management (20%)
- Develop project plans within an interdisciplinary team, deliver allocated tasks, and communicate progress regularly (10%)
- Devise algorithms and logic for new modular, composable software systems, applying industry practices and open-source community standards (5%)
- Enable collaborators to design, implement, reproduce, and extend complex data analyses on real-world research problems, working directly with domain scientists (5%)
- Set technical requirements and standards, and prepare user and developer documentation (5%)
- Stay current on agentic AI for science — models and tools, agentic architectures, evaluation methods — plus security and accessibility practices (5%)
- Serve as technical lead for one or two software development projects (5%)
- Advance the work of the UC Berkeley Open Source Program Office by demonstrating the value and impact of open-source practices (5%)
Required qualifications
- Advanced programming skills, including Python expertise
- Demonstrated fluency using agentic AI and LLM coding agents (such as Claude and/or Codex) for software engineering
- Experience building software, tooling, or infrastructure for agentic AI / LLM systems — agent frameworks, tool and skill integrations, or evaluation harnesses
- Strong software development and maintenance practice: git, testing, CI/CD (GitHub Actions or similar), secure software development
- Experience reviewing code openly on GitHub, with a track record of issues/PR-driven, transparent project management
- A curious mindset and appetite for navigating complexity and uncertainty; a track record of scoping and delivering on requirements from interdisciplinary teams
- Advanced skills in software specification, design, implementation, and deployment at large scale
- Excellent project leadership, communication, and complex problem-solving skills
- Bachelor's degree in a related area and/or equivalent experience/training
Preferred qualifications
- Experience with HPC and scientific workflow systems: job schedulers (SLURM), containerization (Docker, Apptainer/Singularity, Kubernetes), workflow tools (Snakemake, CWL, Dask, Nextflow)
- Experience participating in multi-stakeholder open-source communities: distributed decision-making, consensus-building, coordinating paid and volunteer contributors, mentoring
- A track record building tools used by research teams in the life sciences, physical sciences, geosciences, social sciences, humanities, or public health
- Experience conducting research in a data-intensive scientific domain
Applications are handled by UC Berkeley: the Apply button takes you to the official UC Berkeley posting.
About Lightcone Research
Lightcone Research is an open-source initiative building the foundations of AI-assisted science that is reproducible, composable, and verifiable. We are hosted, in France, by the AISSAI Center (AI for Science and Science for AI, UAR2036) of the French National Centre for Scientific Research (CNRS) and, in the United States, by the Berkeley Institute for Data Science (BIDS) at UC Berkeley. This project is made possible by the support of Eric and Wendy Schmidt.
The AISSAI Center is the CNRS entity dedicated to AI for science and science for AI, with a mission to structure and coordinate cross-cutting initiatives involving every scientific discipline at the interface with AI, from fundamental physics to biology to the humanities and social sciences. BIDS plays a similar role at UC Berkeley: a central hub for data science and open scientific software, connected across the entire campus, from the physical and life sciences to the humanities and social sciences. This dual anchoring places us at the heart of exchanges with a wide range of academic communities and gives us privileged access to the open-source communities based in Berkeley (from Jupyter to Scientific Python).
You will join a small team (fewer than ten people) where your voice will count and where you can contribute actively at every level, from technical solutions to team culture. As one of the first members of the technical team, you will have wide latitude to innovate, propose ideas, and shape our collective adventure.