About Anthropic
Anthropic’s mission is to create reliable, interpretable, and steerable AI systems. We want AI to be safe and beneficial for our users and for society as a whole. Our team is a quickly growing group of committed researchers, engineers, policy experts, and business leaders working together to build beneficial AI systems.
About The Team
Safeguards, Dangerous Asymmetric Harms is a team responsible for developing comprehensive safety systems and policy boundaries across CBRNE (Chemical, Biological, Radiological, Nuclear, and Explosive), Cyber, and Dangerous Asymmetric Advanced Technologies—addressing threats from everyday trust and safety risks to catastrophic AI scenarios. We blend domain expertise in CBRNE and Cyber with ML engineering to create classifiers, evaluation infrastructure, threat models, and conduct RL experiments. The team also performs AI capability uplift testing through partnerships with government laboratories and private industry, leveraging real-world cross-functional experience.
We are looking for a Research Engineer/Scientist, Cyber who can execute rapidly, maintain high throughput, and bring a strong builder mindset to solving complex problems. The ideal candidate will combine deep cybersecurity domain expertise with advanced ML capabilities to build systems that evaluate and prevent dangerous capability development. You'll be designing novel approaches to detect threats that span from traditional cyber attacks to AI-enabled offensive operations, requiring both technical sophistication and creative problem-solving.
This role is primarily focused on building advanced ML systems x Cybersecurity. You will use your deep technical expertise to inform ML solutions that prevent real world harm.
Responsibilities
- Apply ML/AI research to build evaluation systems for cybersecurity safety, with focus on attack pattern characterization and threat detection
- Design and train specialized AI models for cyber threat classification, leveraging network data, exploit analysis, and attack protocols
- Work with SG research to design and implement state-of-the-art ML approaches for identifying dual-use cyber capabilities and zero-day exploits
- Build systems that target the transition of digital cyber capabilities into real-world attacks, preventing malicious exploitation
- Create and implement technical systems for monitoring emerging cyber threats and AI-enabled attack vectors
- Develop classifiers that can distinguish between legitimate security research and potential offensive cyber operations
- Build sophisticated evaluation infrastructure for measuring AI capability uplift in cybersecurity domains
- Design adversarial testing frameworks that probe model capabilities in cybersecurity contexts
- Integrate cyber range validation data with ML training pipelines to improve classifier accuracy
- Develop and maintain cyber threat datasets and benchmarks while ensuring appropriate information security
- Create tools that allow cybersecurity experts to quickly develop and deploy new threat detection evaluations
- Write production-quality Python code for high-throughput security data processing and evaluation systems
- Contribute to cyber risk assessments that directly inform AI model release decisions and policy development
- Work cross-functionally with cybersecurity policy experts, security researchers, and ML engineering teams
You may be a good fit if you
- You are a creative asymmetric hacker who can solve complex and highly technical problems across fields
- Have deep domain expertise in cybersecurity, particularly in offensive security, vulnerability research, or cyber defense
- Possess hands-on keyboard offensive security experience including penetration testing, red teaming, or exploit development
- Have worked with cyber threat characterization, risk assessment, or cybersecurity policy development
- Demonstrate experience in applying ML to security problems, such as malware analysis or intrusion detection
- Have familiarity with ICS/SCADA systems and critical infrastructure security
- Possess security certifications such as SANS certifications, OSCP, or similar credentials
- Can bridge technical cybersecurity knowledge with ML/AI applications for security purposes
- Have experience fine-tuning large language models for specialized domains
- Understand the intersection of converging technologies (AI, autonomous systems, cyber-physical systems) and security risks
- Possess strong foundation in both security techniques and modern ML frameworks (PyTorch/TensorFlow)
- Have experience translating complex technical findings into policy recommendations
- Demonstrate ability to work with sensitive information while maintaining appropriate security protocols
- Show experience with government cybersecurity programs or national security applications
- Can operate effectively in fast-paced environments while maintaining technical rigor
- Have published research in relevant cybersecurity or AI security domains
Do not rule yourself out if you do not fit every qualification - we recognize that the intersection of advanced cybersecurity and ML for security applications is a rare combination. If you have deep expertise in cyber threats and are eager to apply ML to prevent catastrophic risks, please consider applying.
What Makes This Role Unique
- Mission-critical impact: Your work will directly prevent the development and deployment of AI-enabled cyber weapons
- Unique technical intersection: Combine cutting-edge ML with deep cybersecurity domain expertise in ways that have never been done before
- Cross-domain innovation: Apply lessons from traditional cybersecurity to emerging AI threats while pioneering new approaches
- Real-world validation: Collaborate with government agencies and security laboratories to ensure practical applicability
- Dual-use navigation: Balance the advancement of beneficial security research with the prevention of malicious applications
Key Attributes
- Hacker mentality - relentlessly motivated to find gaps
- Curious and creative - approaches problems from unexpected angles
- Dependable under pressure - manages tight deadlines without dropping balls
- Technical depth with policy awareness - bridges both domains effectively
- Self-sufficient builder - can create and run evaluations without engineering support
Representative Projects
- Build infrastructure for running large-scale model evaluations across multiple risk domains
- Create tools for rapid evaluation prototyping and iteration
- Contribute to evaluation frameworks that could become industry standards
- Design and implement custom testing environments for specific capability assessments
- Develop monitoring and analysis systems for evaluation results
- Collaborate with domain experts to translate theoretical risks into practical tests, such as cyber ranges and autonomous replication environments
Candidates Need Not Have
- Domain expertise in all specific risk areas
- 100% of the skills needed to perform the job
- Prior experience with AI model evaluation
Annual Salary
The expected salary range for this position is:
$280,000 - $340,000 USD
Logistics
Education requirements: We require at least a Bachelor's degree in a related field or equivalent experience.
Location-based hybrid policy: Currently, we expect all staff to be in one of our offices at least 25% of the time. However, some roles may require more time in our offices.
Visa sponsorship: We do sponsor visas! However, we aren't able to successfully sponsor visas for every role and every candidate. But if we make you an offer, we will make every reasonable effort to get you a visa, and we retain an immigration lawyer to help with this.
We encourage you to apply even if you do not believe you meet every single qualification. Not all strong candidates will meet every single qualification as listed. Research shows that people who identify as being from underrepresented groups are more prone to experiencing imposter syndrome and doubting the strength of their candidacy, so we urge you not to exclude yourself prematurely and to submit an application if you're interested in this work. We think AI systems like the ones we're building have enormous social and ethical implications. We think this makes representation even more important, and we strive to include a range of diverse perspectives on our team.
How We're Different
We believe that the highest-impact AI research will be big science. At Anthropic we work as a single cohesive team on just a few large-scale research efforts. And we value impact — advancing our long-term goals of steerable, trustworthy AI — rather than work on smaller and more specific puzzles. We view AI research as an empirical science, which has as much in common with physics and biology as with traditional efforts in computer science. We're an extremely collaborative group, and we host frequent research discussions to ensure that we are pursuing the highest-impact work at any given time. As such, we greatly value communication skills.
The easiest way to understand our research directions is to read our recent research. This research continues many of the directions our team worked on prior to Anthropic, including: GPT-3, Circuit-Based Interpretability, Multimodal Neurons, Scaling Laws, AI & Compute, Concrete Problems in AI Safety, and Learning from Human Preferences.
Come work with us!
Anthropic is a public benefit corporation headquartered in San Francisco. We offer competitive compensation and benefits, optional equity donation matching, generous vacation and parental leave, flexible working hours, and a lovely office space in which to collaborate with colleagues.