Reality Defender (YC W22)

• Design, build, and optimize ML/DL models for production-scale audio deepfake detection, ensuring robustness across diverse real-world conditions including compression artifacts, noise, telephony, and streaming pipelines. • Partner with clients to develop a deep understanding of their production environments and define model performance criteria. • Investigate failure cases in client environments, build custom evaluation frameworks, and implement mitigation strategies spanning both Engineering and AI. • Design and execute structured experimentation roadmaps aligned with client requirements and proactive system resilience goals. Translate findings into clear and actionable insights. • Monitor, measure, and report on model performance in production using data analytics and AI observability tools (e.g. Datadog, Metabase). Identify degradation trends, data drift, and emerging threat patterns before they impact client outcomes. • Build and maintain dashboards and analytics pipelines that surface model health metrics, enabling data-driven decisions across AI, Engineering, and Product teams. • Collaborate with cross-functional partners — Applied Scientists, Deployment Engineers, and Product teams — to deploy scalable, production-grade models with clear performance benchmarks and monitoring in place.

• Tune and optimize ML/DL models for production-scale audio deepfake detection. • Investigate failure cases in the client environment, build custom evaluation frameworks, and implement mitigation strategies for model robustness. • Drive model iteration for performance under a variety of real-world environments, e.g. compression artifacts, noise, telephony, and streaming pipelines. • Present technical findings and model performance insights to internal stakeholders. • Interface with Product and Engineering teams to build a deep understanding of the production environment and incorporate relevant evaluations for performance assessment.

• Work with ML engineers and researchers to ensure delivered datasets are usable and correctly scoped • Build and run batch data generation and preprocessing jobs for image, video, and audio data • Execute preprocessing pipelines using shared batch orchestration tools • Design and run ETL jobs to ingest, transform, and organize data in our warehouse. • Validate input and output datasets (schema, metadata, basic quality checks) • Collect, organize, and deliver processed datasets using established conventions • Support creation of development and prototype datasets ahead of large-scale backfills • Maintain version control of data processing repositories following industry best practices • Debug data pipeline failures, ETL issues, and data quality problems

• Tune and optimize ML/DL models for production-scale audio deepfake detection. • Investigate failure cases in the client environment, build custom evaluation frameworks, and implement mitigation strategies for model robustness. • Drive model iteration for performance under a variety of real-world environments, e.g. compression artifacts, noise, telephony, and streaming pipelines. • Present technical findings and model performance insights to internal stakeholders. • Interface with Product and Engineering teams to build a deep understanding of the production environment and incorporate relevant evaluations for performance assessment.

• Design, build, and optimize ML/DL models for production-scale audio deepfake detection. • Interface with clients to build a deep understanding of the production environment and scope out model performance criteria. • Investigate failure cases in the client environment, build custom evaluation frameworks, and implement mitigation strategies that span across Engineering and AI. • Drive model tuning and iteration for performance under a variety of real-world environments, e.g. compression artifacts, noise, telephony, and streaming pipelines. • Design and execute structured experimentation roadmaps, centered around client requirements, anticipated future attack vectors, and proactive system resilience. • Collaborate with cross-functional partners (Applied Scientists, Deployment Engineers, Product teams) to deploy robust and scalable models. • Present technical findings and model performance insights to clients as well as internal stakeholders.

• Investigate and propose new methods for detecting generative multi-modal content, spanning audio and vision modalities. • Perform research on multi-modal deepfake detection and reasoning tasks. • Collaborate with researchers in the team. • Write up results of research for internal reports and submission to academic journals/workshops. • Independently implement and evaluate ideas on modern deep learning stack - Python, PyTorch, and GPU-enabled cloud compute, like AWS/GCP.

• Investigate new feature extraction and data augmentation techniques for generative image/video detection. • Collaborate with scientists and engineers across the organization. • Perform research into deepfake image/video detection. • Aid in integrating insights gained from research into RD products. • Write up results of research for internal reports and submission to academic journals/workshops.

• Define and implement the overall QA strategy, processes, and standards across the organization • Lead test planning, execution, and release validation for platform features, APIs, and AI/ML-powered detection systems • Design, develop, and maintain automated test frameworks for functional, integration, regression, and performance testing • Partner with engineering and product teams to ensure quality is integrated throughout the software development lifecycle • Identify, track, and drive resolution of defects, performance issues, and reliability risks • Establish and monitor quality metrics, test coverage, and release readiness criteria • Mentor and support QA engineers while helping scale the QA function as the company grows • Contribute to continuous improvement of CI/CD pipelines, automation, and release processes

• Design and implement end-to-end product features across frontend, backend, and infrastructure. • Build scalable APIs and backend services that power real-time deepfake detection. • Create intuitive, high-performance UIs that make complex AI outputs understandable and actionable. • Collaborate closely with ML engineers, product managers, and designers. • Own technical decisions and contribute to system architecture. • Improve performance, reliability, and security across the stack. • Mentor other engineers and help define engineering best practices.

• Architect and operate high-availability, low-latency cloud infrastructure for real-time AI inference services. • Build and maintain CI/CD pipelines that support rapid, safe iteration across backend and ML systems. • Own production reliability: monitoring, alerting, incident response, and capacity planning. • Lead infrastructure automation using Infrastructure as Code (Terraform, Pulumi, etc.). • Partner closely with backend and ML engineers to productionize new models and services. • Harden our systems for security, compliance, and data protection. • Optimize cost, performance, and scalability as usage grows. • Establish best practices around observability, deployment strategies, and disaster recovery.