Senior Machine Learning Engineer/ Research Scientist

About The Team

The AI & Biometrics team at Tools for Humanity owns the machine learning behind the World Network. Our iris and face recognition systems, anti-spoofing pipeline, and models running on the Orb enable Proof of Personhood at scale. The team deploys models on dedicated hardware, cross-validates against an expanding identity set via anonymized multi‑party computation, and prioritizes first‑time right results to avoid regressions across millions of verified humans.

Bitte vergewissern Sie sich, dass Sie über das richtige Maß an Erfahrung und Qualifikationen verfügen, indem Sie den vollständigen Überblick über diese Stelle unten lesen.

We are structured around sub‑teams that own the end‑to‑end pipeline: Mobile handles on‑device deployment, ML Infrastructure and MLOps manage data pipelines, training clusters, GPU fleet, and tooling. Researchers and engineers focus on modeling, while infrastructure handles data, compute, and deployment internally.

The team collaborates closely with Orb software, Proof of Personhood, and Product to take ideas from research notebooks to deployed binaries.

About The Role

We are looking for a Senior or Staff Machine Learning Engineer / Research Scientist to push our biometric systems beyond current performance and security benchmarks. This senior IC role operates independently on hard, open‑ended problems, raises team standards, and owns work from research through production.

In This Role, You Will

• Improve core identification and anti‑spoofing models, training and iterating on deep‑learning architectures, losses, and data pipelines with size, latency, and memory constraints as primary design constraints.

• Apply classical computer vision and image processing where appropriate, and favor lighter solutions on‑device when beneficial.

• Lead independent research initiatives: formulate hypotheses, design ablations, run experiments, and determine the right time to ship.

• Diagnose data failures: pull misclassified samples, generate hypotheses, and iterate quickly.

• Build evaluation and monitoring pipelines that detect regressions before production and surface data drift early.

• Take prototypes through testing to deployment, partnering across teams and translating between ML, embedded, and secure‑compute constraints.

• Write design docs, experiment write‑ups, and proposals that endure and drive alignment on contentious decisions.

• Help shape technical standards across the AI & Biometrics team—evaluation methodology, experimentation discipline, model versioning, monitoring—and mentor junior researchers and engineers as a default behavior.

You Might Thrive In This Role If You Have

• An “in‑the‑driver’s‑seat” operating style with ownership of end‑to‑end problems.

• Strong fundamentals in classical computer vision and image processing—OpenCV, NumPy, filters, transforms, morphology, geometric methods.

• Hands‑on experience training and shipping deep‑learning computer‑vision models for production quality under tight latency and memory constraints.

• A pragmatic, applied‑research mindset that balances rigor and shipping schedule.

• Solid mathematical fluency for spotting pathologies without running experiments.

• Experimental discipline—written hypotheses, targeted ablations, and judicious judgment on results.

• A collaborative operating style: mentoring, knowledge sharing, and constructive engagement across teams.

• Strong plus: Direct experience with biometric identification at scale, margin‑based metric learning losses (ArcFace, Triplet, and variants), anti‑spoofing / presentation attack detection, adversarial evaluation of ML systems, and publications at top ML venues.

Additional Nice‑to‑haves

• Hands‑on experience with Rust for high‑performance code paths and speed optimization.

• Experience with edge optimization and on‑device deployment—quantization, pruning, distillation, kernel‑level optimization, deployment to mobile NPUs, embedded GPUs, microcontrollers, or other constrained targets. xayajpt

• A background in sensors, imaging, computational photography, or camera ISPs.

..