As the landscape of technology evolves rapidly, senior software engineers find themselves in challenging positions: developing AI architectures without established patterns, modernizing systems under intense delivery pressure, and making nuanced long-term decisions amid technological flux. The acceleration of AI adoption outpaces the availability of best practices, highlighting the urgent need for validated patterns sourced from real-world production systems rather than experimental frameworks. The critical questions loom large: *How do you construct reliable AI infrastructure at scale? What strategies can de-risk agentic systems? And what does a production Ready, Actionable, and Generative (RAG) pipeline look like when catering to 100 million users?*
The upcoming QCon San Francisco 2025, scheduled for November 17–21, offers an invaluable opportunity for practitioners who have navigated these complex issues. This conference will showcase sessions rooted in practical experiences, featuring architectural decisions, implementation challenges, and lessons learned by teams from powerhouse companies such as Netflix, Meta, Intuit, and Anthropic.
AI Sessions: Implementation Over Theory
Hien Luu, a member of the QCon Program Committee, emphasizes the focus of this year’s event: “Senior engineers don’t need another talk on ‘what is GenAI.’ They’re seeking answers to pressing questions like, ‘How do I build production-grade infrastructure for this?’ and ‘How do I validate my architecture for an agentic system?’ This conference will provide insights from professionals who have successfully tackled these issues.”
Featured Topics at QCon San Francisco 2025
- GenAI Infrastructure at Scale: Maggie Hu and Merrin Kurian from Intuit will dissect their production stack, including vector stores, prompt management, and RAG pipelines, demonstrating how they handle infrastructure for 100 million users.
- Agentic System Architecture: Adam Wolff from Anthropic will share insights on architectural tradeoffs made while building the first agentically accelerated software project, focusing on choosing speed over complexity.
- Scaling AI Organizational Capabilities: Amit Navindgi will provide a practical blueprint filled with design patterns to enhance AI capabilities and developer productivity for Zoox Intelligence.
- Post-Training Techniques for Production LLMs: Faye Zhang and Andi Partovi will detail techniques that enable large language models (LLMs) to function effectively in production environments.
- AI Platforms for Reliability: Aaron Erickson from NVIDIA will explore platform designs that balance deterministic tools and exploratory agents, ensuring reliability in diverse operational contexts.
- Reinforcement Learning for Ad Generation: Alex Nikulkov from Meta will discuss how they utilized reinforcement learning to enhance the generation of ad text, showcasing the practical application of this AI technique.
- Evaluating AI Copilots for Productivity: Sepehr Khosravi from Coinbase will provide frameworks for assessing and selecting AI copilots, essential for standardizing tools effectively within teams.
- Explore the full AI schedule for more insights and topics.
Beyond AI: Architecture, Platform Engineering, and Reliability
Platform Engineering & CI/CD
- CI Green Across Multiple Language Monorepos: Dhruva Juloori from Uber will share how MergeQueue maintains a stable mainline while processing hundreds of changes across various programming languages, offering applicable solutions for developers in polyglot environments.
- Automating Software Changes at Scale: Casey Bleifer from Netflix will delve into maintaining secure and updated software across diverse environments, presenting automated fleet management techniques that prevent disruption.
Architecture & Scalability
- Building Mission-Critical Platforms: Matthew Liste, with over 20 years at American Express, will present lessons learned from developing resilient infrastructures, including insights on balancing security, stability, and scalability.
- Centralized Data Deletion Platform at Netflix Scale: Vidhya Arvind and Shawn Liu will share their experiences in orchestrating a centralized data deletion platform, covering crucial aspects like observability and trade-offs in a live traffic context.
- Modernizing Relevance Systems for a Billion Users: Nishant Lakshmikanth from LinkedIn will recount their transition from offline batch systems to real-time recommendations, detailing migration phases that resulted in a dramatic reduction in compute costs while enhancing user engagement.
Security in AI Development
- Securing AI-Accelerated Development: Sriram Madapusi Vasudevan from AWS Agentic AI will cover critical elements of threat-modeling AI pipelines, focusing on safeguarding sensitive data and strategies for validating AI-generated code.
- For a comprehensive exploration, view the full conference schedule.
Mark your calendars: QCon San Francisco 2025 runs from November 17–21. Don’t miss out on the early bird pricing ending on November 11. For more information and to register, visit qconsf.com.
Inspired by: Source
