Senior Machine Learning Engineer

Xident; Union (NJ); Sur place; USD 140 000 - 190 000; Plein temps

Résumé du poste

Xident B.V. seeks a Senior Machine Learning Engineer to advance age-bracket classifiers and on-device face recognition models. You will enable client-side inference via ONNX and WebAssembly to enhance privacy while maintaining performance.

Join a remote-first team, collaborating with Python and Go engineers on training pipelines and model deployment. This role emphasizes fairness, bias mitigation, and state-of-the-art CV techniques like ArcFace.

Qualifications

• 7+ years of machine learning experience with focus on computer vision.
• Deep expertise in face recognition architectures (ArcFace, CosFace, or similar).
• Strong PyTorch skills with experience exporting to ONNX for edge deployment.
• Published research or significant contributions to CV/ML projects.
• Experience with model optimization (quantization, pruning, distillation) for edge inference.
• Understanding of ML fairness, bias detection, and mitigation strategies.

Responsabilités

• Improve our age bracket binary classifiers — currently +18 achieves 0.03% FPR / 11% FRR, target similar for +12/+15/+21/+25
• Research and implement state-of-the-art face recognition using InsightFace (ArcFace) architectures
• Develop robust liveness detection and anti-spoofing models resistant to photos, videos, and deepfakes
• Optimize models for client-side ONNX Runtime inference (WebAssembly, Core ML, TensorFlow Lite)
• Build fair and unbiased models — measure and mitigate demographic performance gaps
• Collaborate with Python and Go engineers on training pipelines and model deployment via River queue

Connaissances

Machine learning

Computer vision

Face recognition

PyTorch

ONNX

Fairness & bias detection

Outils

ArcFace/CosFace architectures

ONNX Runtime

WebAssembly

River queue

Core ML

TensorFlow Lite

Description du poste

Senior Machine Learning Engineer

We're looking for a Senior ML Engineer to advance our age bracket classifiers and face recognition models. We run 5 binary classifiers (+12/+15/+18/+21/+25) deployed as ONNX models for client-side inference via WebAssembly — you'll improve accuracy, reduce bias, and tackle anti-spoofing.

Xident is building the future of digital identity verification. Our platform enables businesses to verify users' ages and identities in seconds, while respecting privacy through our unique "Verify Once, Access Everywhere" model.

Founded in 2023, we've already processed over 10M+ verifications processed and serve 500+ businesses. We're a remote-first company with team members across 12 countries.

Our mission is simple: To make identity verification seamless, private, and accessible for everyone. We believe privacy and convenience shouldn't be mutually exclusive, and we're proving it every day.

25+

Team members

500+ businesses

Customers

12

Countries

What You'll Do

• Improve our age bracket binary classifiers — currently +18 achieves 0.03% FPR / 11% FRR, target similar for +12/+15/+21/+25
• Research and implement state-of-the-art face recognition using InsightFace (ArcFace) architectures
• Develop robust liveness detection and anti-spoofing models resistant to photos, videos, and deepfakes
• Optimize models for client-side ONNX Runtime inference (WebAssembly, Core ML, TensorFlow Lite)
• Build fair and unbiased models — measure and mitigate demographic performance gaps
• Collaborate with Python and Go engineers on training pipelines and model deployment via River queue

What We're Looking For

• 7+ years of machine learning experience with focus on computer vision
• Deep expertise in face recognition architectures (ArcFace, CosFace, or similar)
• Strong PyTorch skills with experience exporting to ONNX for edge deployment
• Published research or significant contributions to CV/ML projects
• Experience with model optimization (quantization, pruning, distillation) for edge inference
• Understanding of ML fairness, bias detection, and mitigation strategies

Nice to Have

• PhD in Computer Science, ML, or related field
• Experience with document OCR and VLM-based verification
• Background in privacy-preserving ML — on-device inference, no server-side face storage

Our Values

We iterate quickly, gather feedback, and improve continuously.

Ownership Mentality

Everyone owns their domain end-to-end. No finger-pointing, just solutions.

Transparent by Default

We're happy to answer any questions before you apply. Reach out to our team anytime.