About Helpurr
Helpurr is building AI-powered tools to help cat owners and veterinarians detect pain earlier. Cats are notoriously difficult to read — they hide pain instinctively, and by the time symptoms are obvious, conditions have often progressed. Our mobile app uses computer vision to analyze cat facial expressions and flag potential signs of discomfort, giving owners an objective tool to support timely veterinary care.
We're a small, early-stage team based in Hong Kong, working at the intersection of veterinary science and machine learning.
The Role
We're looking for a Data Scientist to own and improve the models at the core of Helpurr. You'll work across the full ML lifecycle — from training data pipelines and model experimentation to validation with veterinary professionals and eventual production deployment.
This is a high-ownership role. You won't be handed a spec and asked to implement it — you'll be the person figuring out what to try next, running experiments, and deciding what ships. If you care about doing rigorous applied ML work on a problem that genuinely matters, this is a good fit.
What You'll Do
Improve computer vision models for cat facial analysis, including object detection and keypoint estimation
Improve classification models for pain detection from facial features
Explore and benchmark alternative architectures — propose, prototype, and evaluate new approaches
Own the training data pipeline — design label collection workflows, manage data versioning, and validate data quality
Apply modern ML techniques such as transfer learning, data augmentation, and semi-supervised learning to maximize model performance
Collaborate with veterinarians to validate model outputs against clinical assessments and refine labeling protocols
Analyze labeler agreement and label quality to ensure training data integrity
Deploy models for production inference — optimize for mobile and API serving (this will grow over time as models mature)
What We're Looking For
3-5 years of experience in machine learning or data science, with hands-on model development (not just analysis)
Strong computer vision fundamentals — experience with detection, classification, or keypoint estimation tasks
Proficiency in Python and at least one deep learning framework (PyTorch or TensorFlow)
Experience with transfer learning and working with domain-specific or limited datasets
Comfort with ambiguity — you can define your own experiments, interpret results critically, and iterate without heavy direction
Good engineering habits — version control, reproducible experiments, clean code. You don't need to be a software engineer, but your work should be maintainable.
Nice to Have
Experience with mobile model optimization (quantization, pruning, on-device inference)
Familiarity with MLOps tooling — experiment tracking (W&B, MLflow), model registries, CI for ML
Background in animal behavior, biomedical imaging, or veterinary science
Data engineering skills — comfortable with ETL pipelines, data validation, and cloud infrastructure
Experience with active learning or human-in-the-loop ML workflows
What We Offer
Remote-first — work from anywhere. We're based in Hong Kong and prefer overlap with East Asian timezones (UTC+8, with flexibility of +/- 2-3 hours), but we're open to discussing arrangements for the right person.
Full-time position with competitive compensation
Real impact on animal welfare — the models you build will directly help cats get care sooner
Early-stage ownership — small team, no bureaucracy, direct influence on product and technical direction
Collaboration with veterinary professionals — your work is grounded in clinical science, not just metrics
How to Apply
Send your resume and a brief note about a model or ML project you're proud of. Links to papers, repos, or write-ups are welcome. We care more about what you've actually built than where you went to school.