Research Scientist II

About the role

Pindrop is the Real Human + Right Human® Identity Trust Platform for the AI era. As AI-driven fraud and deepfakes erode trust in digital communication, Pindrop delivers continuous identity verification and deepfake detection across voice, video, and digital interactions in real time. Enterprises rely on Pindrop to secure billions of high-risk customer interactions each year, including top U.S. banks, as well as leading insurers and healthcare providers. Powered by models trained on more than 1.5 billion real-world interactions annually and protected by 300+ patents, Pindrop restores trust while reducing fraud, lowering operational costs, and improving customer experience. Recognized by TIME as one of the Top 10 Most Influential Software Companies of 2026 and by Inc. for Best in Business for Innovation, Pindrop is backed by leading investors including Andreessen Horowitz, IVP, and CapitalG.

Responsibilities

• Build and improve fraud risk models and scoring systems using a combination of audio, behavioral, and metadata-based signals.
• Analyze fraud patterns across customer environments and translate findings into measurable improvements in model performance, investigation workflows, or mitigation strategies.
• Research and build a scam detection stack, from conception to realization.
• Partner with engineering and cross-functional teams to move successful research into production and improve fraud outcomes in live environments.
• Support high-priority fraud investigations by analyzing system behavior, fraudster attack patterns, and detection gaps, then recommending practical next steps for our customers.
• Improve the quality and precision of fraud-related identity signals, including voice-based indicators and repeat-offender detection.
• Design and maintain reproducible research workflows, internal tools, and evaluation pipelines that help the team experiment efficiently and measure impact clearly.
• Contribute to technical reviews, knowledge sharing, and research documentation that helps the broader organization understand and apply your work.
• Contribute to adjacent innovation areas, including emerging AI-assisted fraud-analysis workflows, when relevant to team priorities.

Requirements

• You are persistent, curious, and scientifically rigorous, especially when working through ambiguous data, noisy signals, or fast-evolving fraud behavior.
• You are comfortable owning research workstreams from problem definition through experimentation, analysis, and recommendation.
• You communicate clearly with both technical and non-technical partners, and you can explain tradeoffs, assumptions, and results in a practical way.
• You care deeply about reproducibility, documentation, and building research that can stand up in real production settings.
• You are motivated by high-impact security and fraud problems and want your work to influence real customer outcomes.
• Your skill-set
• Advanced Degree (Master's or PhD) in Computer Science, Mathematics, Statistics, Engineering, Artificial Intelligence, or a related quantitative field, or equivalent applied research experience.
• 3+ years of professional experience in machine learning, large-language models, fraud detection, natural language processing, risk modeling, speech or signal processing, anomaly detection, or a closely related domain.
• Strong Python skills and experience building research tooling, experimentation frameworks, or model evaluation workflows.
• Hands-on experience with modern machine learning frameworks such as PyTorch, TensorFlow, or Keras.
• A track record of translating research findings into practical improvements, whether in models, decision systems, or production-facing recommendations.
• Foundational knowledge of fraud, identity, consumer scams, authentication, risk scoring, or customer security concepts.
• Experience working on fraud or scam detection in voice, IVR, contact center, authentication, or adjacent trust and safety environments.
• Experience working on building and/or fine-tuning multi-modal foundation models.
• Experience improving precision and recall in real-world detection systems, including thresholding, scoring, watchlists, or entity-resolution style signals.
• Familiarity with metadata-driven risk signals such as telephony, carrier, device, account, or behavioral indicators.
• Experience with sequence modeling, event-based risk modeling, or other approaches used to detect evolving attack behavior.
• Familiarity with LLM-enabled research workflows, retrieval systems, or observability tools used to support analys

Benefits

Your Match

How well this role fits your profile.

Company Intel

What employees say

Worked at pindropsecurity? Share your experience