Research Scientist II – Deepfake Detection & Authentication (Video & Audio)

Who We Are Pindrop is a cybersecurity company delivering continuous identity verification across voice, video, and digital interactions. Powered by the Pindrop Intelligence Network, our unified trust platform helps enterprises answer three critical questions: Is this identity synthetic? Detect AI-generated and manipulated voice and video. Is this identity risky? Identify fraud, suspicious behavior, and other risk signals. Is this identity known? Authenticate identities using voice, device, behavioral, and other signals. Pindrop brings this intelligence together across purpose-built trust engines to help organizations make informed decisions in real time — protecting customers, workforce interactions, and emerging AI-driven interactions. Pindrop protects billions of high-risk interactions for enterprises, including leading financial institutions, insurers, and healthcare organizations. Its technology is informed by more than 8 billion real-world interactions annually and protected by 300+ patents. Recognized by TIME as one of the 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. What you’ll do As a Research Scientist II on our core Research team, you will independently drive research initiatives, deliver reproducible experimental results, and help translate machine learning models into real-world product solutions across deepfake video, deepfake audio, and voice/identity authentication . In this role, you will: Conduct research to develop new functionalities and improve existing technologies for synthetic media detection, multimodal deepfake identification (audio and video), and real-time identity authentication . Design, implement, and evaluate deep learning algorithms related to audio-visual deepfake detection, speech synthesis detection, facial manipulation analysis, and biometric authentication . Design, develop, and maintain internal research packages and pipelines to train, evaluate, and benchmark ML models, ensuring computational efficiency and absolute reproducibility . Partner closely with research engineers and engineering teams to transition complex research prototypes into efficient, scalable, production-ready systems across cloud and edge platforms . Support targeted investigations by analyzing model behavior, edge cases, attack vectors, and detection gaps, turning insights into measurable improvements in model robustness . Participate in regular code and research reviews, cross-functional project syncs, and contribute to patents and peer-reviewed papers for top-tier computer vision, speech, audio, and machine learning conferences . Who you are You remain focused, persistent, and scientifically rigorous through multiple experimental iterations, navigating ambiguity in complex multimodal datasets, evolving attack techniques, and shifting methodologies . You seamlessly bridge the gap between research and engineering teams, explaining complex AI/ML and biometric concepts with clarity to both technical and non-technical stakeholders . You take individual ownership of assigned research projects and workstreams, making sound technical decisions grounded in empirical data, mathematical theory, and rigorous experimental validation . You follow through on your commitments, communicate research progress transparently, and maintain the highest standard of technical integrity, documentation, and model reproducibility . You design, train, and evaluate machine learning models with a strong awareness of bias, drift, and operational deployment constraints, applying sound judgment to balance experimentation speed with model quality and robustness . Your skill-set Must-Haves: Master’s degree or PhD in Computer Science, Artificial Intelligence, Electrical Engineering, Mathematics, or a related quantitative field (or equivalent industry research experience) . 3+ years of professional or post-graduate research experience in Deep Learning, with a dedicated focus on at least one core area: computer vision (facial analysis, video processing), speech/audio processing (speaker recognition, speech synthesis detection), or voice biometrics and identity authentication . Hands-on research experience in audio-visual deepfake detection, synthetic media identification, or multimodal biometric verification . Strong programming proficiency in Python with a proven ability to build, maintain, and structure reproducible research tooling, experimentation frameworks, or evaluation workflows . Hands-on mastery of modern machine learning frameworks such as PyTorch, TensorFlow, or Keras . A proven track record of successful, timely project delivery and the ability to translate research findings into practical improvements, supported by peer-reviewed publications or patent contributions . Nice-to-Haves: Functional programming experience in C/C++ or familiarity with model optimization and deployment tools (e.g., ONNX, TensorRT, TF-Lite, Go) . Practical experience with real-time streaming video or audio processing pipelines and large-scale synthetic data generation/augmentation . Experience competing in or leveraging benchmarks from prominent challenges (e.g., DFDC, ASVspoof, NIST SRE/FRVT, Kaggle) . Familiarity with natural language processing (NLP), foundational multimodal architectures, or sequence modeling applied to authentication and spoofing detection . What’s in it for you At Pindrop, research sits at the intersection of AI, voice, video, fraud prevention, and identity trust . You will turn real-world interaction data into insights that protect millions of people from synthetic media and sophisticated identity manipulation : Build expertise in an emerging category: Develop firsthand knowledge of multimodal deepfake detection, generative AI defenses, and real-time authentication as these technologies rapidly evolve . Work with real-world data at meaningful scale: Train and evaluate models on complex audio and video datasets, identifying subtle artifacts, anomalies, and evasion tactics in high-stakes environme