QCon AI New York 2026 Program Committee Highlights
Expert committee selects key sessions on agentic systems, inference, and engineering evolution.
1Conference Scope and Focus
QCon AI New York 2026 takes place December 15-16 at the Westin Jersey City Newport. The two-day event targets senior software engineers, architects, and engineering leaders. Early-bird pricing is set at $1,705 until November 10, increasing to $1,790 thereafter. The program emphasizes production-grade AI implementation and evolving architectural standards for modern teams.
2Agent Identity and Authorization
Nancy Wang, CTO at 1Password, will keynote on identity models for autonomous agents. As agents perform more independent tasks, traditional security frameworks face challenges. The session explores delegated authority and credential management, specifically how agents access necessary secrets across subagent chains without exposing sensitive data within their operational context during production.
3Managing LLM Context Costs
Tejas Chopra of Headroom Labs addresses the constraints of token costs and latency in production systems. The session focuses on compression, caching, and retrieval strategies. Key topics include learned prompt compression and prefix-cache-aware alignment. These techniques aim to mitigate failure modes that frequently impact tool calling and overall agent trajectory performance.
4Evolving Engineering Organizations
Peter Bell, founder of Gather.dev, will examine how coding agents reshape engineering structures. Drawing on insights from companies like Anthropic, OpenAI, and Coinbase, the talk covers shifts in team rituals, hiring practices, and leadership roles. It specifically addresses the changing professional landscape for junior developers as autonomous coding agents become standard.
5Open Source Agentic Inference
Francisco Javier Arceo from Baseten discusses open-source inference using vLLM and the Agentic API project. The session tracks a coding task from harness to tool execution. It investigates where tools run and how infrastructure supports repeated inference. The analysis highlights why cost per completed task is a critical metric beyond token pricing.
6Standardizing Agentic Interoperability
Kevin Kelley from Anthropic will navigate the current landscape of agentic standards including MCP, A2A, and OAuth. The session aims to establish interoperable identity and trust boundaries without introducing new security vulnerabilities. It distinguishes between mature, shipped standards and ongoing research, helping architects decide which frameworks are ready for production implementation.
7Advancements in Audio AI
Tara Sainath of Google DeepMind will keynote on LLM-driven speech and audio modeling. The presentation covers unified multilingual representations and the trade-offs between end-to-end and cascaded architectures. It explores real-time conversational intelligence, detailing Google’s latest research and product developments across speech recognition, generation, and translation tasks for modern enterprise applications.
8Strategic Architectural Implications
The selected sessions reflect a shift toward managing the complexity of autonomous agent ecosystems. By focusing on identity, cost-efficient inference, and organizational structure, the conference addresses the practical hurdles of scaling AI. Architects must prioritize these foundational capabilities to ensure that agentic delivery remains secure, performant, and sustainable within their engineering environments.
Original article: InfoQ — QCon AI New York 2026 Program Committee Highlights