FirstPrinciples Research Fellow
In short AI
Research Fellow fellowship of 6 to 12 months, full-time or equivalent, fully remote, at a non-profit organization building an AI system for physics research.
Description
a long ad (5,001 characters): the AI summarised it; the full text is in the next tab.
- Role
- Research and test model architectures that combine scientific literature, NLP and symbolic reasoning, including representations for physical concepts and mathematical objects and alternatives to transformer-based architectures. Build data ingestion pipelines for scientific literature and benchmarks measuring physical understanding and mathematical reasoning. Manage model training and compute budget, design sandbox environments for autonomous exploration, and implement tests and safeguards. Work with engineers and present regular updates tied to defined milestones.
- Requirements
- PhD or postdoctoral research in Computer Science, Machine Learning, Theoretical Physics or a closely related field
- Research track record in model architectures, representation learning, reasoning systems, or mathematical, physical or formal reasoning
- Demonstrated ability to translate abstract theory into testable computational systems
- Ability to work independently and own a problem end to end
- Preferred
- experience with cross-disciplinary research
- Conditions
- Contract: fellowship of 6 to 12 months, full-time or equivalent
- Salary: fixed stipend for the full term, amount not stated
- Hours: full-time or equivalent, remote
- Location: remote
- Benefits
- none stated
Overview:
FirstPrinciples is a non-profit organization building an autonomous AI Physicist to understand the nature of reality: the underlying structure, governing principles, and fundamental laws of our universe. We're developing an intelligent system that can explore theoretical frameworks, reason across disciplines, and generate novel insights to tackle the deepest unsolved problems in physics. By combining AI, symbolic reasoning, and autonomous research capabilities, we're developing a platform that goes beyond analyzing existing knowledge to actively contribute to physics research. Our goal is to accelerate progress on the questions that have captivated humanity for centuries.
We operate as a global nonprofit organization, with a Canadian foundation, a US-based 501(c)(3).
As part of this effort, we are launching a Research Fellowship Program for advanced AI and physics researchers. Fellows will work directly with the FirstPrinciples Research and Engineering teams to design, test, and implement state-of-the-art (SOTA) methods and applications that will be integrated into the core AI Physicist system. This is not a paper-only fellowship and your work will go straight into production, helping shape how scientific research is performed.
What You will DoAs a Research Fellow, you will own a well-scoped research direction that contributes directly to the AI Physicist's ability to reason about physics.
Core Objectives:
- Improve Theo’s ability to produce scientifically sound, high-quality outputs in 2026;
- Introduce new ideas, methods, and approaches that meaningfully shift system performance;
- Bring deep domain expertise (PhD+ level) in one or more targeted research areas.
Research Areas & Responsibilities
Fundamental Model Research:
- Research, design, and test novel model architectures that combine academic literature, NLP, symbolic reasoning, and structured scientific workflows.
- Prototype and build embedding representations for physical concepts, mathematical objects, and logical structures, enabling models to reason over equations, abstractions, and scientific constraints rather than surface text alone.
- Investigate alternatives to transformer-based architectures and deliver concrete recommendations.
- Design and run targeted experiments to evaluate new architectural ideas, using empirical results to guide the development of next-generation model architectures.
- Develop reinforcement learning loops that enable models to run internal and independent thought experiment.
Multimodal Data & Benchmarking:
- Design and automate scalable data ingestion pipelines that aggregate scientific literature, metadata, equations, and experimental data.
- Create custom benchmarks to measure physical understanding, mathematical reasoning, and failure modes in scientific reasoning and abstraction.
- Refine and release curated datasets and baselines once internal validation is complete.
Training, Testing & Safety:
- Run and track model training jobs while managing compute usage and budget constraints.
- Design sandbox environments for controlled autonomous exploration.
- Build evaluation frameworks using visual and statistical tools to identify strengths and blind spots.
- Implement tests and guardrails that flag low-quality or unsafe outputs.
- Maintain internal issue tracking with clear failure modes and fixes.
Collaboration & Technical Guidance:
- Work closely with engineers to ensure research is feasible and production-ready.
- Communicate technical trade-offs clearly to non-technical stakeholders.
- Present regular research updates tied to defined milestones.
Who This Fellowship Is ForWe are looking for researchers who:
- Can operate independently and own a problem end-to-end.
- Are motivated by ambitious, long-horizon goals.
- Have strong builder instincts, not just theory.
- Value rigor, clarity, and intellectual honesty.
- Are comfortable engaging with exploratory, incomplete results.
Typical profiles include:
- PhD or postdoctoral researchers in Computer Science, Machine Learning, Theoretical Physics, or a closely related field.
- Track record of research in either:
- model architectures, representation learning, or reasoning systems (CS/ML path); or
- mathematical, physical, or formal reasoning applied to fundamental problems (physics path).
- Demonstrated ability to translate abstract theory into testable computational systems.
- Comfortable working across disciplinary boundaries.
Program Structure
- 6-12 month engagement with full-time or equivalent commitment
- Fully remote
- Fellows participate in projects and are expected to produce outputs (code, models, and other deliverables)
- Fixed stipend for the full term
How to Express Interest:If this resonates, we welcome an expression of interest, including:
- Resume;
- Description of your research background;
- Primary areas of expertise; and
- Relevant recent work.
Join us at FirstPrinciples and be a part of a transformative journey where science drives progress and unlocks the potential of humanity.
Originally posted on Himalayas
How to apply
Submit an expression of interest including: resume; description of research background; primary areas of expertise; relevant recent work. Deadline and contact not stated. AI
Prepare your application
A cover letter and a short pitch for this job, written from your profile, and what to learn next.
Sign in and create your profile from your CV. It is private: we share it with no one.
Something wrong or unlawful in this ad? Report it
Similar jobs
- AI ResearcherFirstPrinciples
- Physicist, Theo Collaborators ProgramFirstPrinciples
- Member of Technical Staff, ResearchFirstPrinciples
- Member of Technical Staff, Staff Physicist, Quantum Information and AIFirstPrinciples
- Principal Research Scientist (Computer Science / Computational Physics)micro1
- Master's Student fellowships - "NextGenAI: Center for Responsible AI"Champalimaud Foundation · Lisboa
Training for this job
- ISTQB® Testing with Generative AIGalileuonline · daytime · 14 h · 1,320 € + IVA · starts 09/12/2026Asked for in this ad: AI / machine learning
- Utilizar Ferramentas da Ia na OrganizaçãoServiço de Formação Profissional de Santarémonline · daytime · 25 h · free · starts 02/12/2026Asked for in this ad: AI / machine learning
- a Inteligência Artificial no QuotidianoServiço de Formação Profissional de Viana do Casteloonline · evenings · 8 h · free · starts 03/12/2026Asked for in this ad: AI / machine learning
Languages AI
Published on (2)
- Himalayas seen 16 Nov
- FirstPrinciples seen 17 Nov – 22 Nov
First seen on 16 Nov, 23:01 · last seen on 22 Nov, 21:28