Data Nomads Meetup #3: The Leap From Model to System

​There is a considerable gap between experimenting with machine learning models and bringing a large-scale machine learning system to life. Many challenges arise that were considerably out-of-scope during the model experimentation stage: ensuring system reliability and scalability, managing data pipelines, integrating with existing infrastructure, and making sure the system is maintainable.

During our third meetup, we will focus on designing architectures for large-scale machine learning systems. Speakers with diverse backgrounds—software architecture, LLM-based systems, end-to-end MLOps—will share their approaches to building scalable, maintainable, and reliable systems.

​Speakers & Topics:

1. Nik Sergievskii - 𝘗𝘳𝘢𝘤𝘵𝘪𝘤𝘢𝘭 𝘐𝘯𝘵𝘳𝘰𝘥𝘶𝘤𝘵𝘪𝘰𝘯 𝘵𝘰 𝘚𝘰𝘧𝘵𝘸𝘢𝘳𝘦 𝘈𝘳𝘤𝘩𝘪𝘵𝘦𝘤𝘵𝘶𝘳𝘦
Discover why software architecture is crucial, and learn how to approach it from various perspectives and metrics.

2. Pavel Filipovich - 𝘉𝘶𝘪𝘭𝘥𝘪𝘯𝘨 𝘢 𝘙𝘈𝘎 𝘚𝘺𝘴𝘵𝘦𝘮 𝘪𝘯 𝘵𝘩𝘦 𝘞𝘪𝘭𝘥
Uncover the lessons learned and key (surprising) decisions that can help you avoid pitfalls when building RAG systems.

3. Mikita Shchutski - 𝘔𝘢𝘤𝘩𝘪𝘯𝘦 𝘓𝘦𝘢𝘳𝘯𝘪𝘯𝘨 𝘚𝘺𝘴𝘵𝘦𝘮 𝘋𝘦𝘴𝘪𝘨𝘯 - 𝘌6+ 𝘐𝘯𝘵𝘦𝘳𝘷𝘪𝘦𝘸 𝘌𝘹𝘱𝘦𝘳𝘪𝘦𝘯𝘤𝘦
Detailed outline for ML System Designs with a focus on product metrics, data constraints, and modeling trade-offs, bypassing traditional elements like balancers, caches and queues.

Key takeaways:

• ​Understanding the architectural components of machine learning systems.

• ​Best practices for deploying and maintaining ML models in production.

• ​Real-world case studies showcasing successful model-to-system transitions

​Language: English