ml engineer

Voodoo creates mobile games and apps, including games such as Mob Control and Block Jam and apps such as BeReal and Wizz. Its Engineering & Data team builds technology products and platforms for its gaming and consumer apps. The Voodoo Ad-Network builds advertising network services and uses first-party data to optimize and scale monetization.

• Take end-to-end ownership of projects from ideation to production release, including scoping features, estimating timelines, designing architecture, and benchmarking technologies
• Build and maintain data and ML pipelines; optimize training pipelines for performance, memory efficiency, and cost
• Provide reusable, tested pipeline components for Data Scientists and review contributions to shared code
• Extend dataset capabilities by integrating data sources, scaling feature windows, and increasing training data volumes within pipeline constraints
• Contribute to deep learning development, including GPU workload orchestration, custom PyTorch training loops, and model architecture support
• Maintain reproducibility and consistency across the ML lifecycle through versioned configs, experiment tracking, and online-offline consistency tooling
• Collaborate with infrastructure teams on scalability, node pools, resource monitoring, and CI/CD migrations
• Participate in a weekly rotation to triage and resolve alerts from Airflow, dbt, and related systems
• Collaborate daily with back-end developers, data scientists, infrastructure engineers, and product managers
• Share knowledge and contribute to an engineering culture where team members feel supported and empowered to grow

• At least 5 years of experience as an ML Engineer or in a similar role
• Experience with end-to-end ML ownership, including problem framing, baselines, experimentation, deployment, and iteration
• Extensive Python knowledge for ML pipeline code, including preprocessing, training and evaluation workflows, experiment utilities, and reproducible configs
• Ability to implement training mechanics, including custom steps and metrics, data-loading patterns, and performance-conscious preprocessing
• Practical experience training and evaluating models with scikit-learn, LightGBM, and PyTorch
• Experience optimizing memory usage during data preprocessing and model training
• Experience with hyperparameter tuning, experiment tracking, and reproducible training using configs, seeds, and versioning
• Experience with Amazon Web Services and familiarity with scalability, reliability, and security
• Understanding of production ML challenges, including feature stores and training-serving skew
• Excellent English communication skills
• Будет плюсом: experience with model-serving infrastructure, including real-time or batch inference and latency or throughput optimization

• Benefits depend on the country of residence