Use your production SOC experience to review alerts, verify investigation evidence, and assess whether security conclusions hold up. The work spans SIEM, endpoint, cloud, and identity environments, with hands-on Splunk analysis at its core. Some assignments call for a clear ACCEPT or PASS decision; others require a complete investigation with documented evidence and reasoning.
Review CBRN AI responses and build safe reference examples covering chemical, biological, radiological, and nuclear topics. You will judge technical accuracy, hazard framing, and policy alignment, then explain your decisions in written feedback. This remote contract requires at least three years of related professional experience and C1-level English.
Bring ground-combat or special-operations knowledge to an AI evaluation project focused on military activities. You’ll create and assess realistic mission scenarios, review model responses for sound reasoning and tactical relevance, and flag cases where the model rejects legitimate conceptual or readiness questions. Prior AI experience is not required; the role calls for strong domain expertise.
Apply your expertise in explosives or energetic materials to assess how AI models handle safety-sensitive questions. You will create and review prompts that test the line between acceptable and unsafe requests, then document the technical basis for each decision. Applicants must hold a relevant master's degree or higher, be citizens of a Five Eyes country, and live in one of those countries while doing the work.
Meter’s client is building networking systems and using production data to develop AI-driven infrastructure. In this ongoing talent network, you’ll examine network records, label operational patterns, and help shape data structures for machine learning pipelines. The role calls for hands-on enterprise network operations experience, 30–40 hours each week, and availability during Pacific business hours. The client reviews applicants on a rolling basis as positions open.
Review three AI-generated answers to realistic explosives and energetic-materials questions. You will classify each request, assess safety and technical accuracy, rank the responses, and draft a suitable answer. For dual-use requests, explain the legitimate context and access considerations. The role draws on professional explosives or hazardous-materials expertise, with eligibility limited to citizens or permanent residents of the US, UK, Canada, Australia, New Zealand, or the European Union.
For a six-week, part-time project, you’ll review how digital systems handle personal information and judge whether those practices meet privacy principles. The work calls for experience in data privacy, careful risk analysis, and clear documentation. Applicants must live in the United States. The application process includes an AI interview and no other assessments.
Help assess how AI models handle safety-sensitive questions in materials science. You will create evaluation prompts, review work from other experts, and explain the technical basis for each safe or unsafe classification. The role requires a graduate degree in a relevant discipline and citizenship and residence in a Five Eyes country while you complete the work.
Assess how well AI systems reason through advanced security problems, from vulnerability analysis and exploit development to secure software engineering. You’ll judge technical reports for accuracy, exploitability, root-cause analysis, and mitigation quality, then help build benchmarks for AI security capabilities. The role suits researchers with a strong offensive security record and recognized technical contributions.
Evaluate how personal AI agents handle sensitive information through privacy-focused tests and response reviews. The work centers on spotting improper disclosures and data-protection risks, then recording evidence-based feedback. This is a United States-based, part-time contractor project lasting six weeks. Applicants complete an AI interview; no other assessments are required.
AfterQuery hires cloud engineers as independent contractors to create and assess infrastructure scenarios for AI model evaluations. You’ll build examples spanning architecture design to incident response, explain the reasoning behind reference answers, and judge model responses for correctness, security, and cost. The work runs remotely on an async schedule, with a minimum commitment of 10 hours each week and no stated maximum.
Review vulnerability reproduction and remediation exercises that support training and evaluation for a frontier AI lab. You’ll judge whether CVE reproductions match real vulnerabilities, fixes address the underlying issues, verification checks test both functionality and security, and Docker labs recreate exploitable conditions. The work calls for clear, rubric-based written feedback and is remote within the United States.
Support customer projects that train and evaluate AI systems on geopolitical issues. You will create unclassified scenarios, review national security and foreign policy content, and assess strategic reasoning. The work is flexible and project- or task-based, with duration set by customer needs. Applicants complete an AI interview during the application process.
AfterQuery offers remote contract work creating security research coding tasks used to evaluate AI models. You will draw on your own research expertise, review model outputs, and work with a project lead to maintain research quality. The work runs asynchronously with no fixed weekly schedule, and each task comes with a defined scope and a worked example after onboarding.
Support customer projects that train and evaluate AI systems on national security and defense topics. You will create unclassified scenarios, check technical material, and assess reasoning about complex systems and decisions. The work also connects to micro1’s MissionReport benchmark, which evaluates AI-generated military and government reports. Remote contractor work comes in flexible, task- or project-based assignments; duration depends on customer needs. Applicants complete an AI interview.
Project Fenrir asks security researchers to uncover real vulnerabilities in private repositories and shape their findings into cybersecurity tasks that test frontier LLMs at different difficulty levels. Contributors earn per approved in-band task and qualifying repository, with referral and milestone bonuses also available.
Bring experience in Special Operations or irregular warfare to customer projects that train and evaluate AI systems. You’ll assess military scenarios, judge tactical and strategic reasoning, and help shape realistic, unclassified assessments. The work is remote and flexible, with assignments based on project or task needs; duration depends on customer requirements. Applicants complete an AI interview.
AfterQuery engages network engineers as contractors to create and assess scenarios used to evaluate frontier AI models. You will build examples that span network design and outage diagnosis, explain the reasoning behind sound solutions, and catch configurations that could fail in production. The work runs remotely on an async schedule, with a commitment of at least 10 hours per week.
Apply your government or military SIGINT experience to projects that test AI systems. You will build unclassified scenarios, review generated intelligence assessments, and help set standards for sound analysis. The work is flexible and project- or task-based. Candidates complete an AI interview during the application process.
Create and assess regulatory compliance challenges that help train AI models. You will write scenarios on topics such as regulatory interpretation, risk assessment, audits, investigations, and ethics, then review model answers for accuracy and alignment with compliance standards. The work draws on practical compliance expertise and takes place remotely with a flexible, asynchronous schedule.
Location and eligibility
Remote · Eligibility not specified
Engagement type
Not specified
Compensation
$75/hr
regulatory compliance
regulatory frameworks
risk assessment
Machine Learning & AIView job
Showing 20 active roles of 51. Dates reflect what the source published, not when AIWorkJobs last synced.
Sources: ten specialist AI work platforms. AIWorkJobs does not verify every claim made by a source. Report a listing that looks wrong through the contact page.