AI Safety Argumentation Platform Research Engineer
The case that AGI and ASI pose catastrophic risks is strong but poorly systematized: fragmented across literatures, inconsistently formalized, and vulnerable to motivated dismissal. CARMA is building an evidentiary infrastructure to fix this. It combines ontologies, knowledge graphs, defeasible argumentation frameworks, and LLM-assisted population pipelines under expert curation, feeding structured argument content into communications flows that reach policymakers, technical audiences, journalists, and the public.
In this role, you'll develop and operate that system. You'll work where argumentation theory meets agentic AI tooling, building machinery that is both formally tractable and persuasive in practice, the epistemic backbone that will help stakeholders elucidate why good arguments for prospective expectations are good, and why bad arguments are bad.
This position is 100% remote but requires occasional travel.
About CARMA
The Center for AI Risk Management & Alignment (CARMA) works to help society navigate the complex and potentially catastrophic risks arising from increasingly powerful AI systems. Our mission is specifically to lower the risks to humanity and the biosphere from transformative AI.
We focus on grounding AI risk management in rigorous analysis, developing policy frameworks that squarely address AGI, advancing technical safety approaches, and fostering global perspectives on durable safety. Through these complementary approaches, CARMA aims to provide critical support to society for managing the outsized risks from advanced AI before they materialize.
CARMA is a fiscally-sponsored project of Social & Environmental Entrepreneurs, Inc., a 501(c)(3) nonprofit public benefit corporation.
Responsibilities
Extend ontologies and knowledge graph schemas representing claims, evidence, argument structures, defeaters, and confidence
Implement defeasible argumentation frameworks (e.g., ASPIC+, Dung-style, argumentation schemes) that capture both logical structure and vulnerability to rebuttal
Operate and quality-control LLM-driven population pipelines, with cross-check scaffolds, provenance tracking, and human-in-the-loop curation
Architect agent coordination patterns for multi-step research and population tasks, with robust error handling and graceful degradation
Pre-harden argument structures by mapping the strongest counterarguments, steel-manned objections, and known defeaters
Build export pipelines that translate structured argumentation into diverse communications formats across audiences and registers
Maintain current awareness across AI safety, capabilities, and governance sufficient to know when new developments require graph updates, and to know where to find authoritative further detail
Collaborate with communications staff and researchers to ensure outputs serve real persuasive needs
Required Qualifications
Working familiarity with formal or semi-formal argumentation theory (abstract or structured argumentation, defeasible reasoning, dialectical models, or argumentation schemes)
Experience with ontology engineering or knowledge graph development (OWL/RDF, property graphs, or equivalent)
Operational experience with LLM agent systems: agent coordination platforms, prompt engineering at scale, and QC regimes for LLM outputs (adversarial probing, consistency checks, calibration)
Fluent vibecoding practice: rapid prototyping and shipping with LLM-assisted development in production-adjacent contexts
Substantive grounding in AI safety, AI governance, and current frontier-AI dynamics, with the literacy to locate authoritative sources on any sub-topic or human expertise in the space
Familiarity with philosophy of science concepts bearing on evidence: defeaters, burden of proof, inference to the best explanation, underdetermination
Good coding skills; comfort with graph databases or query languages
Experience designing cross-check and verification scaffolds for unreliable automated processes
Sound judgment about when a claim is well-supported versus when it needs hedging, further substantiation, or withdrawal
Self-directed; strong written communication
Preferred Qualifications
Graduate work or equivalent depth in argumentation theory, computational argumentation, epistemology, or philosophy of science
Familiarity with AIF, Carneades, or comparable computational argumentation tools
Track record in AI safety or governance (publications, policy work, or substantive community contributions)
Background in argument mining, claim extraction, or stance detection
Experience with debate formats or structured deliberation methods
Understanding of motivated reasoning, belief change, and cognitive biases as they bear on communications strategy
Open-source contributions in any relevant area
CARMA/SEE is proud to be an Equal Opportunity Employer. We will not discriminate on the basis of race, ethnicity, sex, age, religion, gender reassignment, partnership status, maternity, or sexual orientation. We are, by policy and action, an inclusive organization and actively promote equal opportunities for all humans with the right mix of talent, knowledge, skills, attitude, and potential, so hiring is only based on individual merit for the job. Our organization operates through a fiscal sponsor whose infrastructure only supports persons authorized to work in the U.S. as employees. Candidates outside the U.S. would be engaged as independent contractors with project-focused responsibilities. Note that we are unable to sponsor visas at this time.
\n
\n$160,000 - $210,000 a year
\n
Apply To This Job
In this role, you'll develop and operate that system. You'll work where argumentation theory meets agentic AI tooling, building machinery that is both formally tractable and persuasive in practice, the epistemic backbone that will help stakeholders elucidate why good arguments for prospective expectations are good, and why bad arguments are bad.
This position is 100% remote but requires occasional travel.
About CARMA
The Center for AI Risk Management & Alignment (CARMA) works to help society navigate the complex and potentially catastrophic risks arising from increasingly powerful AI systems. Our mission is specifically to lower the risks to humanity and the biosphere from transformative AI.
We focus on grounding AI risk management in rigorous analysis, developing policy frameworks that squarely address AGI, advancing technical safety approaches, and fostering global perspectives on durable safety. Through these complementary approaches, CARMA aims to provide critical support to society for managing the outsized risks from advanced AI before they materialize.
CARMA is a fiscally-sponsored project of Social & Environmental Entrepreneurs, Inc., a 501(c)(3) nonprofit public benefit corporation.
Responsibilities
Extend ontologies and knowledge graph schemas representing claims, evidence, argument structures, defeaters, and confidence
Implement defeasible argumentation frameworks (e.g., ASPIC+, Dung-style, argumentation schemes) that capture both logical structure and vulnerability to rebuttal
Operate and quality-control LLM-driven population pipelines, with cross-check scaffolds, provenance tracking, and human-in-the-loop curation
Architect agent coordination patterns for multi-step research and population tasks, with robust error handling and graceful degradation
Pre-harden argument structures by mapping the strongest counterarguments, steel-manned objections, and known defeaters
Build export pipelines that translate structured argumentation into diverse communications formats across audiences and registers
Maintain current awareness across AI safety, capabilities, and governance sufficient to know when new developments require graph updates, and to know where to find authoritative further detail
Collaborate with communications staff and researchers to ensure outputs serve real persuasive needs
Required Qualifications
Working familiarity with formal or semi-formal argumentation theory (abstract or structured argumentation, defeasible reasoning, dialectical models, or argumentation schemes)
Experience with ontology engineering or knowledge graph development (OWL/RDF, property graphs, or equivalent)
Operational experience with LLM agent systems: agent coordination platforms, prompt engineering at scale, and QC regimes for LLM outputs (adversarial probing, consistency checks, calibration)
Fluent vibecoding practice: rapid prototyping and shipping with LLM-assisted development in production-adjacent contexts
Substantive grounding in AI safety, AI governance, and current frontier-AI dynamics, with the literacy to locate authoritative sources on any sub-topic or human expertise in the space
Familiarity with philosophy of science concepts bearing on evidence: defeaters, burden of proof, inference to the best explanation, underdetermination
Good coding skills; comfort with graph databases or query languages
Experience designing cross-check and verification scaffolds for unreliable automated processes
Sound judgment about when a claim is well-supported versus when it needs hedging, further substantiation, or withdrawal
Self-directed; strong written communication
Preferred Qualifications
Graduate work or equivalent depth in argumentation theory, computational argumentation, epistemology, or philosophy of science
Familiarity with AIF, Carneades, or comparable computational argumentation tools
Track record in AI safety or governance (publications, policy work, or substantive community contributions)
Background in argument mining, claim extraction, or stance detection
Experience with debate formats or structured deliberation methods
Understanding of motivated reasoning, belief change, and cognitive biases as they bear on communications strategy
Open-source contributions in any relevant area
CARMA/SEE is proud to be an Equal Opportunity Employer. We will not discriminate on the basis of race, ethnicity, sex, age, religion, gender reassignment, partnership status, maternity, or sexual orientation. We are, by policy and action, an inclusive organization and actively promote equal opportunities for all humans with the right mix of talent, knowledge, skills, attitude, and potential, so hiring is only based on individual merit for the job. Our organization operates through a fiscal sponsor whose infrastructure only supports persons authorized to work in the U.S. as employees. Candidates outside the U.S. would be engaged as independent contractors with project-focused responsibilities. Note that we are unable to sponsor visas at this time.
\n
\n$160,000 - $210,000 a year
\n
Apply To This Job