QCon AI New York 2026 · December 15–16 · New York
Code that arrives faster than it can be reviewed, rising token costs, evals you can act on, and agents that need permissions and guardrails. The senior engineers running these systems show you how they're handling them.
Need approval? Get the email for your managerConference
December 15–16, 2026
Location
New York
Format
Curated by senior practitioners
Past attendees included senior engineers from
Why attend
Senior engineers and architects share what works on systems they run and what they'd do differently, so you can test your own plans against it.
A look at how teams outside your own company are solving the same problems, such as agent permissions and eval design.
Hundreds of senior engineers attend, with one speaker for every eleven. There's time to work problems through with peers and ask speakers questions 1:1.
The Program Committee invites only senior practitioners leading the way or doing something new. Sponsors do not control the program. No hidden product pitches.
Program
How senior engineers are handling the challenges of AI in production.
Keynote
Nancy Wang
CTO @1Password, Founded and Scaled AWS Data Protection
“Identity systems assume a human is behind every credential: one principal, a stable role, a bounded session, and someone accountable afterward. Agents hold none of these.”
01 · Agent architecture and authority
Featured talk
Pratik Agarwal
Distributed Systems Engineer @Figma, Previously @Momento and @AWS DynamoDB
At Figma, the trust boundary for agents extends beyond the sandbox. Warm pools cut provisioning latency by 90%, and availability is up from 99% to 99.95%.
02 · Agents in production operations
Featured talk
Ronak Nathani
Principal Staff Software Engineer, Compute Infra @LinkedIn
LinkedIn runs one operations agent across 500,000+ nodes and 5M+ pods. The guardrails sit on the server, including rate limits, delete protection and peer approval for production changes.
03 · Evaluation and continuous improvement
Featured talk
Fabio Flores
Software Engineer @DoorDash, Leads Foundation Development for Agentic Ordering
DoorDash checks every change to its agentic ordering system before launch, with an LLM judge calibrated against human reviewers. The same rubrics score simulated and real sessions, so gains measured offline carry into production.
04 · Context and model serving
Featured talk
Fan Yang
Staff Software Engineer @Netflix, Building the AI Serving Platform
Rajat Shah
Staff Software Engineer @Netflix, Leads the Technical Architecture for ML Model Serving
Netflix serves about 1 million inference requests per second across 300+ models from one platform, after five years of consolidation. Fan and Rajat explain what multi-tenancy costs to operate.
05 · AI-assisted software engineering
Featured talk
Min Chen
Principal Staff Engineer @LinkedIn, Caltech CS PhD
LinkedIn's AI code review covers tens of thousands of pull requests a week across about 10,000 repositories, with more than 60% of sampled suggestions accepted. A validation pipeline filters findings before developers see them.
06 · Engineering leadership and adoption
Featured talk
Liz Fong-Jones
Technical Fellow @honeycomb.io, Over Two Decades of Experience
Peak weekday PR merges to Honeycomb's main monorepo went from about 30 to about 74 between early 2025 and April 2026, and have held steady since. Liz explains what keeps quality from slipping.
Speakers
Presentations
Explore selected presentations and the practitioners sharing them.
Program Committee
The Program Committee is made up of working senior engineers. They invite every speaker, looking for people who run the system in production and will say what they'd do differently. Sponsors do not control the program, so the talks are vendor-neutral and carry no hidden product pitches.
Timeline
Status: Done
Preliminary schedule
Status: Now
Complete program published
Status: Next
Hear about the latest QCon and InfoQ developments. Be the first to learn about in-person and virtual events.
Operating an autonomous AI system at scale is one of the hardest engineering problems we have right now. QCon AI New York looks entirely at the runtime reality. You get direct access to how teams handle traces, zero-trust boundaries, inference costs, and the organizational workflows that keep these systems from breaking.
our attendees
Senior engineers, architects and engineering leaders making decisions on production AI.
Are senior developers or higher, architects, leads and engineering management
One speaker for every eleven attendees, so the people running these systems are easy to reach between talks.
Software architects
Patterns for where an LLM's judgment belongs and where deterministic code takes over
Staff+ engineers
What it takes to turn a demo agent into a system that survives a failed run
ML and MLOps engineers
Eval designs that say ship or hold
Platform engineers
How other platform teams handle model access, eval pipelines and inference cost
Security engineers
Identity, least privilege and kill switches for tool-using agents
Engineering leaders
Adoption metrics tied to delivery and quality
We offer group discounts for teams of three or more. Email info@newyork.qcon.ai with your team size to get your discount code.
Past QCon attendees included senior engineers from organizations including Amazon, Netflix, Apple, Google, Meta, Microsoft, eBay, Uber, JPMorgan, Goldman Sachs, and NASA.
Why senior engineers trust QCon
Senior engineers, architects and engineering leaders on what they took back from QCon AI New York.
I came in with a question: how much of what's presented at an AI conference is real, and how much is demo with more lights and noise than substance? And today there was a lot of substance, more than I expected.
The strongest message I brought was not about models. It was about execution, going beyond the hype and understanding how companies run AI at full scale.
Being surrounded by engineers and AI practitioners building real, scalable systems gave me a fresh perspective on how thoughtful design, trade-offs, and long-term impact shape good software. What stood out most was the community and the honesty.
Convince your manager
We've added a convince your manager page with an email template and an AI prompt. Both build your request around your current work, the sessions that apply to it, what you'll bring back and the cost.
venue & hotel
Find venue directions and hotel information for QCon AI New York, December 15-16, 2026.
faq
A two-day conference on production AI for senior software engineers, software architects, and engineering leaders. Senior practitioners walk through the AI systems they run in production and what they'd do differently, across agent runtime design, context engineering, evals, tool access, security, and inference cost. December 15–16, 2026, at The Westin Jersey City Newport, one PATH stop from Lower Manhattan.
Senior software engineers, architects, Staff+ engineers, ML and MLOps engineers, platform engineers, security engineers, and engineering leaders making decisions on production AI. If you're moving AI work from prototype into production and own how it runs, it's built for you.
C4Media, the team behind InfoQ and QCon. QCon has helped senior software teams adopt new technologies and practices for more than 20 years, and QCon AI applies the same practitioner-led approach to production AI.
Yes. Register 3 or more colleagues from the same company and save per ticket: $50 off for 3 to 6 attendees, $75 for 7 to 10, $100 for 11 to 15, and $125 for 16 or more. QCon alumni get $100 off. Discounts don't stack and are applied at the time of order. The group registration deadline is November 27, 2026.
Email info@newyork.qcon.ai with your group size for a code.
Registration fees are non-refundable, but you can transfer your ticket to a colleague for free until December 11, 2026. Email info@newyork.qcon.ai to arrange it.
Please include the following information for the new attendee:
By a program committee of working senior engineers. Every session has to clear two questions: has the speaker run this in production, and will they share what didn't work as honestly as what did. The committee filters product pitches out before any session is accepted, and sponsored talks are clearly separated from the editorial program. That's what "no hidden product pitches" means in practice.
First sessions are announced in August, a preliminary schedule follows in October, and the complete program is published by early November. Two reasons to register before then: pricing rises on set deadlines, so booking sooner costs less, and every session comes from the same committee filter, so you're not buying blind even before the names are up.
December 15–16, 2026, at The Westin Jersey City Newport, 479 Washington Boulevard, Jersey City, NJ 07310. The conference runs inside the hotel, one PATH stop from Lower Manhattan.
We've partnered with BCD M&E to help QCon AI NY participants find affordable accommodations that meet their requirements.
Access to a wide selection of hotels at the lowest available rates, with no administration or handling fees. Group discounts available for 9+ rooms.
Two days of talks from senior engineers who build and run AI systems in production. December 15–16, 2026, New York.
RegisterOther QCon events
Take a look at:
QCon AI NY · Dec 15–16
$1,620 until October 13
