Senior software engineers are being asked to design and verify AI architectures without established patterns, modernize systems under delivery pressure, and make long-term technical decisions while core technologies are still evolving. With AI adoption accelerating faster than best practices can emerge, teams need validated patterns from production systems, not experimental frameworks. The gap isn’t conceptual; it’s implementation: How do you actually build reliable AI infrastructure at scale? How do you de-risk agentic systems? What does a production RAG pipeline look like at 100 million users?
QCon San Francisco 2025 (November 17–21) brings together practitioners who’ve already solved these problems in production. The sessions focus on architectural decisions, implementation constraints, and lessons learned from teams at Netflix, Meta, Intuit, Anthropic, and others.
Early bird pricing ends November 11.
AI Sessions: Implementation Over Theory
“Senior engineers don’t need another talk on ‘what is GenAI'”, said Hien Luu, QCon Program Committee member. “They’re asking: ‘How do I build production-grade infrastructure for this? How do I validate my architecture for an agentic system?’ These sessions answer those questions from people who’ve already built it”.
What’s covered:
- GenAI infrastructure at 100M-user scale: Maggie Hu, Group Product Manager for the Core AI Platform, and Merrin Kurian, Distinguished Engineer for AI Foundations at Intuit, break down their production stack: vector stores, prompt management, RAG pipelines.
- Agentic system architecture: Adam Wolff, Engineer on Claude Code at Anthropic, shares the architectural tradeoffs from building the first agentically accelerated software project. Speed over complexity is the theme.
- Scaling AI org capabilities: Amit Navindgi, Staff Software Engineer leading Zoox Intelligence at Zoox, has a practical blueprint with design patterns for scaling AI capabilities and developer productivity.
- Post-training techniques for production LLMs: Faye Zhang (Staff Engineer, GenAI search lead) from Pinterest and Andi Partovi (Co-Founder, Veris AI; former GenAI Solution Architect, Google) walk through the techniques that make LLMs actually work in production.
- AI platforms for reliability: Aaron Erickson, Senior Manager and Founder of DGX Cloud Applied AI Lab at NVIDIA, explores how to design platforms that balance deterministic tools with exploratory agents.
- Reinforcement Learning for generative ad text: Alex Nikulkov, Research Scientist (RL lead for Monetization GenAI) at Meta, shares how they improved ad generation using reinforcement learning.
- Choosing AI copilots for productivity: Sepehr Khosravi, ML Platform Engineer at Coinbase, provides a framework for evaluating and selecting AI copilots. Useful if you’re trying to figure out which tool to standardize on.
- Explore the full AI schedule.
Beyond AI: Architecture, Platform Engineering, Reliability
Platform Engineering & CI/CD
- Keeping CI green across multiple language monorepos: Dhruva Juloori, Senior Software Engineer and core SubmitQueue contributor at Uber, walks through MergeQueue, the CI scheduling system that keeps Uber’s mainline green while processing hundreds of changes per hour across Java, Kotlin, Swift, Go, Python, and more. If you’re dealing with polyglot monorepos, this one’s directly applicable.
- Automating changes across a diverse fleet: Casey Bleifer, Senior Software Engineer at Netflix, covers how to maintain up-to-date and secure software at scale across polyglot environments without disrupting productivity. Automated fleet management without the chaos.
Architecture & Scalability
- Building resilient, mission-critical platforms: Matthew Liste, Head of Infrastructure at American Express, has been building infrastructure at American Express for over 20 years, including electronic trading and credit card processing. His talk distills lessons on sustaining platforms long-term, balancing currency with client disruptions, and maintaining security, stability, and scalability. This kind of long-view perspective is rare.
- Designing a centralized data deletion platform at Netflix scale: Vidhya Arvind, Tech Lead & Founding Architect for Netflix’s Data Abstraction Platform, and Shawn Liu, Senior Software Engineer at Netflix, share how they architected a centralized data deletion platform. The talk covers orchestration, observability, journaling, and the tradeoffs between throughput, safety, and scalability under live traffic.
- Modernizing relevance systems for 1B users: Nishant Lakshmikanth, Engineering Manager at LinkedIn, walks through their multi-year migration from offline batch systems to real-time recommendations serving nearly a billion users. Four migration phases, decoupling candidate generation from scoring, Embedding-Based Retrieval for cold-start, and LLM-powered ranking that cut offline compute costs by 90%+ while boosting engagement.
Security
- Securing AI-accelerated development: Sriram Madapusi Vasudevan from AWS Agentic AI covers threat-modeling AI pipelines, guarding sensitive data, detecting prompt-injection, and validating AI-generated code before it merges. Particularly relevant given how fast teams are shipping AI features.
- View the full conference schedule.
QCon San Francisco 2025 runs November 17–21. Early bird pricing ends November 11. View the full schedule and register at qconsf.com.
