Charm Security is building the first agentic AI workforce for scams and human-centric fraud. Our AI agents work alongside fraud teams at Tier 1 banks in North America: investigating cases, guiding frontline teams through high-risk interactions as they happen, and gathering intelligence beyond the organization's perimeter.
We do it by understanding manipulation, intent, and human behavior, through our proprietary HVE™ (Human Vulnerabilities and Exploits) framework, which combines fraud expertise with behavioral psychology. That's how we take institutions from detection to action.
We're looking for an Applied AI Researcher to build the agents and models behind that platform. This role suits a data scientist or AI researcher who wants their research to ship: someone energized by deploying AI at scale, working under real-time and adversarial constraints, and seeing the impact on people who would otherwise have lost their savings.
What you'll do
Design, build, and evaluate the AI agents at the core of our platform, across fraud investigation, frontline guidance, and threat intelligence.
Model how manipulation actually works: adapt and fine-tune LLMs to recognize intent, deception, and coercion in real conversations.
Build scalable, real-time pipelines that turn behavioral signals, case data, and the alerts our customers already generate into decisions our agents can act on.
Optimize AI models for latency, accuracy, and adaptability in high-volume fraud prevention use cases.
Work AI-natively. Use coding agents and AI tooling to prototype, experiment, and ship faster.
Stay at the forefront of LLMs, ML, and AI research, implementing the latest advancements in those fields.
Requirements
4+ years of experience in Data Science, AI, or NLP, preferably in fraud detection, security, or fintech.
MSc in Computer Science, Machine Learning, Engineering, Mathematics, or a related field.
Strong experience with agentic techniques and ML tooling.
Hands-on experience building, evaluating, and shipping AI agents, not only prototyping them.
Fluent with AI-assisted development (coding agents, LLM tooling) as part of how you actually work day to day.
Experience bringing a product from ideation to production, and owning it end to end.
Comfortable with ambiguity and ownership: early-stage pace, end-to-end responsibility.
Nice to have
Background in cybersecurity, threat intelligence, fraud, or financial crime AI.
Experience with RAG and vector databases (Pinecone, Weaviate, pgvector) and AI-driven search.
Experience working with financial institutions or in regulated, enterprise-scale data environments.
Familiarity with privacy-preserving AI or federated learning.
This is a unique opportunity to be at the forefront of AI-driven fraud prevention, building the agentic workforce that stands between the world's largest banks' customers and the people trying to manipulate them. 🚀