Design, train, evaluate, and deploy computer vision models for face matching (selfie ↔ document / reference image) and face liveness detection / anti-spoofing (detecting attacks from photos, replay, masks, deepfake, etc.).
Optimize models for on-device inference (mobile, edge) — focus on resource constraints (latency, memory, compute).
Build data pipelines: image/video collection, preprocessing, augmentation, alignment, quality checks, normalization.
Monitor model performance in production: metrics like FAR / FRR, ROC / AUC, error cases under challenging conditions (low light, extreme pose, blur).
Collaborate with mobile / SDK / backend teams to integrate models securely into client applications.
Research state-of-the-art techniques in face recognition, spoof detection, domain generalization, and apply to real-world constraints.
Continuously iterate and improve based on real-world feedback, edge cases, and adversarial examples.

Bachelor’s / Master’s / PhD in Computer Science, AI, Electrical Engineering, or related fields.
3+ years of experience in computer vision / face recognition / liveness detection / anti-spoofing.
Proven track record of deploying vision models to production, ideally on mobile / edge platforms.
Strong programming skills in Python; familiarity with frameworks like PyTorch, TensorFlow, ONNX, and mobile inference frameworks.
Experience with face detection, face alignment, landmark detection, preprocessing, and image normalization workflows.
Experience in liveness detection methods: texture / reflection / depth cues / motion cues / multimodal approaches.
Experience optimizing models (quantization, pruning, knowledge distillation) for deployment under resource constraints.
Strong problem solving, debugging, and analytical skills.
Good communication skills; ability to explain trade offs to product, mobile, and operations teams.