AI Engineer: Computer Vision

Company Description Zyphe provides a privacy-first identity verification solution that prioritizes user control over personal data while ensuring businesses are protected from fraud and data breaches. Powered by a decentralized platform, Zyphe enables seamless identity verification and retention without storing Personally Identifiable Information (PII) on company servers. With advanced KYC, AML, and KYB modules built on Web3 principles, Zyphe helps organizations meet modern privacy and security requirements. The platform also offers users secure identity vaults and effortless one-click verification for smooth onboarding experiences. Role Overview We're looking for an AI Engineer specializing in Computer Vision to design and deploy real-time visual intelligence systems at the core of our identity verification platform. This is not a research-only role. You will own the full lifecycle, from model prototyping to production deployment, building the perception layer that powers face detection, liveness checks, and document verification. You'll work at the intersection of deep learning, edge inference, and privacy-preserving AI, shipping models that run fast, accurately, and securely. What You'll Do - Design and train computer vision models for face detection, liveness verification, and document analysis

• Build and optimize real-time inference pipelines using PyTorch, TensorFlow, and ONNX Runtime
• Deploy models to edge and cloud environments with low-latency constraints (TensorRT, CoreML)
• Develop data augmentation and synthetic data strategies to improve model robustness
• Build evaluation frameworks and monitor model drift in production
• Collaborate with backend engineers to integrate CV models into the verification pipeline
• Research and implement privacy-preserving techniques (on-device processing, federated approaches)
• Maintain MLOps infrastructure for training, versioning, and deployment What We're Looking For - Strong experience building and deploying computer vision models in production
• Deep knowledge of CNN and transformer architectures for vision tasks
• Hands-on expertise with PyTorch or TensorFlow and model optimization toolchains
• Experience with edge deployment (ONNX, TensorRT, CoreML, or similar)
• Solid understanding of MLOps practices (experiment tracking, CI/CD for models, monitoring)
• Familiarity with face detection, recognition, or document processing pipelines
• Strong Python skills and comfort with cloud infrastructure (AWS/GCP)
• Ability to reason about privacy implications of visual data processing What Makes You a Great Fit - You think in latency budgets and accuracy tradeoffs, not just loss curves
• You're obsessed with shipping models that work in the real world, not just on benchmarks
• You combine research depth with engineering rigor
• You don't just train models, you own them end-to-end in production