Principal / Staff Data Scientist
@ XsollaPrincipal / Staff Data Scientist
About the job
Xsolla is a global commerce company supporting game developers in funding, distributing, and monetizing games. The role focuses on leading ML initiatives, developing models, and guiding teams to build scalable, production‑ready ML systems in the gaming industry.
Requirements
- Experience with ML model deployment
- Skills in supervised learning
- Knowledge of MLOps tools
- Hands-on model training and tuning
- Experience in fraud detection models
Qualifications
- Advanced Degree in Stats or ML
- Track record in high-impact data science
- Production ML experience
- Familiarity with neural networks
- Knowledge of model monitoring
Full job description
We are looking for an accomplished Principal Machine Learning Engineer to join our global ML organization. In this role, you will drive innovation across our machine learning ecosystem, architect advanced ML solutions, and mentor junior ML engineers around the world. You will play a key part in shaping our technical direction—leading complex ML initiatives, elevating engineering standards, and guiding teams as they build scalable, production‑ready machine learning systems.
If you are ambitious, energized by solving challenging technical problems, passionate about developing talent, and excited to influence the future of ML/AI in the video game industry, this could be the perfect role for you.
Requirements:
Modeling Depth
Advanced Degree in Statistics, machine learning or related areas. Experience in statistics/ML expertise with a track record of leading high-impact data science initiatives at scale of billions of transactions.
Hands-on experience creating, training and fine-tuning models not just integrating hosted model APIs. You should be able to walk through the data, the objective, what broke, and the before/after evaluation numbers, and why the model did not perform as expected.
Experience owning models in production: deployment, monitoring, drift detection, retraining — with real latency budgets, not just research notebooks.
Production experience with classical ML for fraud/anomaly detection, recommendation, or churn/LTV (gradient boosting, deep learning, graph-based models).
Technology Familiarity
Supervised learning, transfer leaning on machine learning as well as neural networks
Basic LLM knowledge, especially how to use it and where not to use it.
MLOps foundations: feature stores, experiment tracking, model registries (MLflow/W&B-class), continuous training pipelines.
Model serving and inference optimization (vLLM-class serving, quantization).
Nice-to-Have:
Publications, conference talks, or recognized open-source contributions to training/eval tooling .
Graph-based fraud detection (fraud rings, device/account linkage)..
Gaming, payments, fraud, advertising domain experience.
Hands-on, up-to-date experience with modern AI tools (e.g., Claude, Copilot, Cursor) for code generation, review, and accelerating day-to-day engineering work.