DeepLearning.AI brings Andrew Ng and Yann LeCun to AI Dev NYC November 30–December 1

DeepLearning.AI’s AI Dev NYC 2026 will take place November 30–December 1, 2026, at Chelsea Piers in New York City. Hosted by the DeepLearning.AI team and Andrew Ng, the two-day technical conference pairs keynotes with workshops on building AI systems. A start time, keynote timezone and official watch link have not been announced on the event page.

Registration is open, with Standard and Pro passes available. The organizer says every pass includes the full agenda, workshops, networking and a post-event recap. Pro adds one year of DeepLearning.AI Pro. Team tickets are also offered, with inquiries directed to aidev@deeplearning.ai.

DeepLearning.AI advertises an audience of more than 2,000 AI engineers and builders. That is the organizer’s attendance expectation, not a confirmed turnout. The conference is presented as a technical gathering for people building with AI, with sessions intended to help attendees choose a technology stack and apply skills to their own projects.

$599Standard

The Standard pass covers the full two-day program and all workshops.

$799Pro

The Pro pass includes the program and workshops, plus one year of DeepLearning.AI Pro.

Ng is scheduled to deliver the opening keynote, The AI Engineering Skills Map. Yann LeCun’s listed talk, World models and what comes after LLMs, looks beyond large language models. Together, those titles give attendees two distinct subjects to follow: the skills needed to build AI systems and the direction of future model development.

Several other listed speakers focus on the work between a model’s capabilities and a deployed system. OpenAI’s Salil Sethi is scheduled to discuss turning new capabilities into reliable production systems. Sonar CEO Tariq Shaukat’s topic is agent-written code at Fortune 500 scale, while Oracle’s Richmond Alake will address memory for agents running in production.

The agent sessions also cover how systems receive context and work together. Unblocked CEO Dennis Pilarinos is listed to discuss where agent context comes from. SandboxAQ’s Timo Bozsolik-Torres will address building agent teams, and Hugging Face’s Joel Niklaus is listed for a two-part Agent Harnesses workshop.

The published session list assigns difficulty levels from beginner to expert. Its subjects span financial workflows, information retrieval, code security, agent memory and deployment. For prospective attendees, those labels offer a way to distinguish introductory architecture work from more advanced implementation topics before choosing a pass.

• Beginner: Architect Agentic Systems for Financial Workflows is the introductory option in the listed program.
• Intermediate: Sessions cover GraphRAG with knowledge graphs, analysis of AI-generated code for quality and security, and browser automation for agents.
• Advanced: The agenda lists persistent memory for agents, repository context for coding agents, and deploying reliable AI agents in production.
• Expert: The program includes meta-harness architectures for AI agents and engineering synthetic data for open large language models.

For readers interested in evaluation rather than deployment, Veris AI co-founders Andi Partovi and Mehdi Jamei are listed for Breaking agents on purpose: the art of better benchmarks. Neo4j’s Nyah Macklin will address finding wrong evidence before users do. These are scheduled teaching topics; the agenda does not identify them as product releases.