The Economy ML team is building the machine learning backbone for Roblox’s virtual economy - from intelligent pricing and personalized storefronts to dynamic layout optimization and avatar content understanding. We partner closely with Marketplace, Payments, and Developer Monetization to deliver ML systems that drive long‑term value across engagement, retention, and creator earnings.

As a Senior ML Software Engineer on the team, you will build the infrastructure and backend services that train, serve, and integrate the models powering ranking, personalization, and pricing across Roblox’s economic surfaces, including Marketplace, Developer Monetization, and Payments. Your work determines how quickly we can take new models from idea to production, and how well they perform once they are serving millions of users.

You Will

• Build and operate the training infrastructure behind Economy ML’s ranking, recommendation, and pricing models: distributed training jobs, efficient data loading, and reproducible pipelines from raw data to trained model.
• Co-design models and systems with ML engineers: profile and optimize training and inference for our ranking and recommendation models, improving GPU utilization, memory, throughput, and latency (e.g., moving models to GPU serving, request batching, mixed precision, faster request-time feature processing).
• Design and build the backend services and APIs that bring model predictions into Marketplace, Payments, and Developer Monetization, with clear latency budgets and graceful fallbacks.
• Build and scale the batch and real-time data and feature pipelines that feed training and serving, with a focus on freshness, correctness, and training/serving consistency.
• Own the path from trained model to production serving: packaging, deployment, versioning, shadow and canary rollouts, and integration with A/B experiments on Roblox’s ML platform.
• Make the systems you own reliable and observable, and set the team’s best practices for logging, monitoring, alerting, and on-call across ML services.
• Partner with Roblox’s central ML platform and data infrastructure teams, and with Marketplace / Economy backend teams, to adopt shared infrastructure and contribute improvements back to it.
• Contribute to the technical direction of the Economy ML team: help shape our training and serving stack, infrastructure roadmap, and the tradeoffs between model quality, latency, and cost.
• Mentor other engineers through design reviews, code reviews, and hands-on guidance, helping raise the technical bar of the broader Economy organization.

You Have

• 5+ years of software engineering experience, including 2+ years building and operating ML infrastructure or ML-backed production services at high scale. Experience with recommendation, search, or ads systems is a strong plus.
• Solid engineering fundamentals in Python and at least one other backend language, with a track record of designing and operating distributed, low-latency services.
• Hands-on experience training and serving deep learning models in production with PyTorch, including scaling training across GPUs and nodes and deploying models on GPU inference servers. Comfortable profiling and optimizing for utilization, memory, latency, and cost.
• Experience building batch and streaming data pipelines and feature stores (e.g., Spark, Flink, Kafka), with attention to data quality and training/serving skew.
• You build systems that are easy to debug and operate, and you take ownership of production health, from monitoring and alerting to incident response.
• You enjoy working closely with ML engineers, platform teams, product managers, and data scientists across Marketplace, Payments, and Developer Monetization, and you can drive alignment in complex, multi‑stakeholder environments.
• You understand how infrastructure choices (latency, freshness, cost) affect model quality and product metrics, and you prioritize work by its impact on users, creators, and the business.
• BS, MS, or Ph.D. in Computer Science, Engineering, or a related technical field, or equivalent practical experience.