NOV 5–6, 2026 · SHANGHAI

AI Engineer Shanghai

AI Engineer came to China for the first time. In Shanghai, the people building open models, agents and robots spent two days together, on stage and off.

Shanghai co-organizers

Mengying Li

Standup Partners

Led international partnerships and worked with speakers.

Yuzheng Sun

Superlinear Academy

Shaped the program, invited speakers and connected US and Chinese developer communities.

02days50+speakers≈1,600attendance over two days, organizers’ estimate

Speakers & hosts

Speaker directory ↗

58 speakers and hosts

Conference program

Full program archive ↗

Open models, agents & infrastructure

09:30-17:50 · UTC+8

One stage in the morning. The community panel starts at 13:00; parallel sessions begin at 13:45.

Morning · Main stage09:30-13:00
  1. 09:35

    Welcome

  2. 09:45

    Opening with swyx · Video

    About

    AI Engineer and Latent Space co-founder swyx opens the conference with a ten-minute video message to the Shanghai audience.

  3. 10:10

    From Models to Working Enterprise Agents

    Topic subject to speaker confirmation

    Abstract

    How models, data, tools and execution boundaries come together in a useful enterprise workflow, and what changes when an agent moves from a demo into ongoing use.

  4. 10:25

    Building Agents: Yesterday, Today, Tomorrow

    Abstract

    The way agents get built has changed fast. Yesterday: chains and RAG, debugged by eyeballing prompts. Today: tool-calling agents in long loops, kept reliable by traces, datasets built from production failures, and evals that gate releases. Tomorrow (the best teams are already on this): telemetry becomes an interface for the agent. Coding agents pull scored traces from your application, cluster failures, propose fixes and evals, and back-test them. Today a human is still in the loop - in the future perhaps not? I'll cover which signals to expose to agents, where humans should stay in control for now, and what changes when agents become the heaviest readers of your traces.

  5. 10:45

    When we can do anything, do we still know what we want?

    Abstract

    AI is narrowing the distance between imagination and reality. But as execution becomes easier, doing more, faster becomes the default. We keep producing, reviewing and keeping up, while losing the time to form ideas of our own. If everyone can create but ends up making similar things, what have we actually liberated? Drawing on building tools at Notion and Cursor, creating ryOS, and leaving San Francisco for Taipei, Ryo explores creativity, judgment and autonomy in the AI era. This is not an argument against speed, but a reconsideration of what deserves acceleration, what needs time, and how tools can give life back to people rather than turn life into another production line.

  6. 10:50

    Break

  7. 11:10
    David Li
    Alex Triplett

    Post-Training at the Frontier: Teaching Open Models to Search and Browse the Web

    Abstract

    In this session, You.com’s engineering team will share how we post-trained Nemotron 3.5 Lightning, an open-weight 30B model, to become a stronger agent for web search and browsing, using You.com’s real-time web search API, NeMo Gym, supervised fine-tuning, and reinforcement learning. We’ll walk through the technical approach that enabled the team to iterate rapidly. It lifted accuracy on BrowseComp, a benchmark for agentic web search and browsing, from 37% to 45%, while making the agent more efficient with fewer tool calls. We will cover topics ranging from dataset preparation, supervised fine-tuning, and reinforcement learning to the agent harness and search infrastructure that made fast iteration possible. We’ll end with a blueprint for post-training and improving LLMs on real-world search tasks as well as a look at what’s next. This work was done in collaboration with NVIDIA and CoreWeave.

  8. 11:25

    Full-Stack Inference Engineering: Latency, Throughput and Cost

    Abstract

    Open-source model inference is becoming a compelling option for enterprises, especially those building AI-native applications. Delivering fast, efficient, and reliable inference requires engineering across the full stack, from CUDA kernels and Kubernetes orchestration to quantization, caching, and disaggregation. This talk explores the key design decisions and emerging techniques shaping production inference systems, with practical insights into balancing latency, throughput, cost, and reliability.

  9. 11:40

    Measure the frontier of AI for real world use

    Abstract

    In this session we will discuss the AI eval products and services Arena provides to help frontier model labs and neo labs measure and improve the performance of their models on the global stage.

  10. 13:00

    Lunch · On your own

Afternoon · Main stage13:00-17:50
  1. 13:40

    Community in the AI era

    • AJWaytoAGI · Founder
    • KojiCrossing · Founder; Venture Partner, Zhenfund
    • Kai Qu42 Chapters · Founder & CEO

    Moderator: 高宁 · Linkloud

    What makes an AI community worth returning to? Four community founders discuss quality, member contributions, and how conversations turn into projects and lasting collaboration.

  2. 13:45

    Break

  3. 14:05

    From Model Intelligence to Agent Harness: Enabling Long-Horizon Agents

    Abstract

    As agents take on longer and more complex tasks, sustaining performance becomes a central engineering challenge. Drawing on Step 5 Preview, StepFun’s latest flagship foundation model, and Step Code, its open-source agent harness, this talk explores how models and harnesses work together to help agents stay on track and deliver reliable results in real-world applications.

  4. 14:25

    Dark Factory to Lobsterville: Intent Engineering for Persistent Multiplayer Agents

    Abstract

    Early 2026 we built the Dark Factory to ship OpenClaw the larget open source agent platform in China and the world, parallel agents running overnight, thousands of commits, pure throughput. It scaled until the output stopped matching what we actually wanted. The bottleneck was intent. Agents produced volume without a durable reason to do any of it. I walk though the city of systems "Lobsterville" we are building at OpenClaw to build OpenClaw. From meat proxies to multiplayer steering and the fundamentals of intent handoff and intent engineering to build the future of AI with agents in a collaborative way.

  5. 14:30

    Break

  6. 14:50

    On-Device AI: Bringing Intelligence Closer to Everyone

    Abstract

    How can useful AI run directly on phones, PCs and other personal devices rather than depending on the cloud? This talk explores memory bottlenecks in on-device models, using early expert routing and flash streaming to reduce the resident memory footprint. It also covers combining millisecond-scale local decisions with larger on-device models, with examples from headphones, robots and smart glasses.

  7. 15:10

    Kubernetes Is Not Your Sandbox

    Abstract

    Teams are reaching for Kubernetes to run agent sandboxes, and it's the wrong tool. Kubernetes is built to keep things alive and hold them in a steady state. A sandbox is born, forked, and killed before any of that machinery catches up. The mismatch compounds because the sandbox keeps gaining requirements without shedding any. In eighteen months it went from a fast code-snippet runner, to a stateful box for long-running agents, to ten thousand ephemeral environments that fork for RL rollouts and die in under a second. It has to be all of those at once, a contradiction set no orchestrator was designed to hold. The cost shows up the moment you measure it. We ran the same 50-action bug-fix trajectory across five stacks and got a 12x spread: 12.9s on the fastest, 161.5s on the slowest. The gap isn't compute, it's lifecycle overhead per action. We name every stack and explain the mechanism behind each number.

  8. 15:30

    AI-Native Work at Manus: One Monorepo, Open Tasks

    Abstract

    Most teams add AI tools to an old workflow. At Manus, we rebuilt our workflow around agents. We keep all our repositories, team rules, and agent skills in one shared workspace, called Manus mono. Mono connects to almost every service in the company, so an agent can write code, run tests, check logs, work with databases, and ship to production from one place. Because of this, anyone in the company can open a PR. Engineers, product managers, designers, and operations staff all fix the bugs they find. On top of this, every task in our team is visible to everyone. This decouples a task from the person who started it. When a task needs information, you send the link to the person who knows, and the task continues after they answer. When a colleague is away, someone else can take over. Open tasks also give us data on how people actually work with AI. In this talk, I share how we built this way of working, what went wrong, and how it changed the way our team collaborates.

  9. 15:35

    Break

  10. 15:55

    Powering open source personal agents with Exa search

    Abstract

    Power open-source personal agents with Exa’s live web search, Places, weather, and research through a single API.

  11. 16:15

    Your Agent's Memory Is Next Quarter's Training Data

    Abstract

    An agent that runs twenty steps doesn't issue one retrieval — it issues dozens, and it writes about as much as it reads. The trace it leaves behind is write-heavy, time-ordered, and queried at three granularities: step, span, trajectory. That's the same shape as an RL rollout store. Not similar to — the same. The model labs hit this two years earlier under different names: semantic dedup, decontamination, hard-negative mining. Agent memory and training data are converging on one set of operators — operational writes and analytical scans over the same data, with no ETL in between. That convergence is why we took Milvus from a vector database to a Vector Lakebase. This talk walks through the workloads that forced the shift, and what the architecture looks like underneath.

  12. 16:35

    Building a reliable, multi-region LLM gateway

    Abstract

    The LLM gateway has become a critical piece of infrastructure at Optiver. Engineers, researchers and agents rely on it daily for a variety of tasks, from analysing large volumes of market news and research to running automated workflows. When the gateway goes down, the impact is expensive and quickly noticed. Keeping LLM traffic running reliably across multiple regions and providers has been challenging, with many of the failures outside our control. In this talk, we share our experience building and scaling a multi-region LLM gateway. We walk through the key design decisions behind protocol translation, scaling, provider failover, observability and multi-region high availability, as well as the tradeoffs and footguns we encountered along the way.

  13. 16:40

    Break

  14. 17:00

    Nobody Planned AI Infra: How AI Builders Pushed a Database Beyond the Database

    Abstract

    AI builders keep asking for capabilities that stretch a database beyond its original boundaries: one database per agent, memory across sessions and state that survives disposable sandboxes. This talk follows the production needs that drove those changes, and the resulting choices around isolation, consistency, recovery and observability.

  15. 17:10

    Break

  16. 17:25

    Optimizing Organization-Wide LLM Inference Cost: Seat License, Pay-per-Token, or Provisioned, and Beyond

    Abstract

    Enterprises buy LLM inference through three structurally different pricing models: credit-based subscriptions, pay-per-token APIs, and self-hosted GPU capacity, with little guidance on which to use at a given volume or what signals it's time to switch. This talk introduces the Efficiency Curve: cost per token under the cheapest procurement mode at each volume, and its crossover points. Using a July-2026 pricing snapshot across three model quality tiers, I show a counter-intuitive finding: the subscription-to-API crossover often never economically materializes; a credit plan's idealized per-token rate can undercut every API price at scale, so the real boundary is a terms-of-service seat limit, not economics. The self-hosting crossover, by contrast, is a genuine price crossover, landing between roughly 360-899M tokens/month depending on tier, and it shifts with output-token share. I walk through the derivation and share a monitoring framework that ties each metric to the procurement decision it should inform, drawn from running AI infrastructure inside a regulated bank.

  17. Rapid Fire · Model systems

    17:25–17:45
    1. 17:35

      Reinforcement Learning at Scale on Agent Substrate: Orchestrating Millions of Sandboxes

      Abstract

      RL post-training for coding and agentic models turns every rollout into a sandbox: a fresh environment with a task image, a few commands, a reward, and teardown. A single training run can need Millions of these per hour across thousands of distinct images, with starts arriving in bursts of hundreds and lifetimes measured in seconds. This talk highlights what we learned running 1M sandboxes using an open-source harness on top of Kubernetes called Substrate, and all the various technological innovations in the stack - warm pooling, image acceleration, low latency scalable storage,microVMs, gVisor, golden snapshot and more. We cover the complete stack orchestration layers that make this work in state of the art Al labs including benchmarks and latency assessment.

    2. 17:45

      Life of a Game Developer in AI era

      Abstract

      AI is systematically absorbing the technical core of game development. Code generation tools now produce gameplay scripts, shaders, and engine plugins from natural-language prompts, while AI pair-programmers handle debugging, refactoring, and unit tests autonomously. Procedural content pipelines driven by generative models build entire levels, NPC behaviors, and quest logic without manual scripting. Automated QA bots playtest builds around the clock, flagging bugs faster than human testers. Even optimization — memory profiling, performance tuning, cross-platform porting — is increasingly delegated to AI agents. The developer's role shifts from writing code to reviewing, orchestrating, and validating machine output. This paper examines which engineering tasks AI has already displaced and what expertise remains irreplaceably human.

  18. 17:50

    Day 1 close

Afternoon · AI applications breakout13:45-17:00
  1. 14:05

    Building a Personal Agent for Growth with YouMind

    Abstract

    Yubo shares how to use YouMind to build a Personal Agent for personal growth.

  2. 14:20

    Trust Your Own Evals, Not the Leaderboard: What 10k+ Internal Agent PRs Taught Us About Adopting Open Models

    Abstract

    Every team picking a model right now is reading someone else's leaderboard. We stopped, and the reason was an accident of how we shipped. Harvey released its cloud agent platform to our own engineers before we shipped long-horizon agents to lawyers. Internally it has now driven 10k+ merged pull requests across our repositories. That internal deployment was not just a safety net — it became the eval set. Once you have thousands of real agent runs on your own workload, you can measure a model against the work you actually do, instead of trusting a public benchmark that was never scored on your tasks. That measurement is what makes it rational to move to an open model, or to fine-tune one, rather than simply hoping it holds up. This talk covers the infrastructure that made that possible, and the four places it broke first: - Durable runs: what happens to a long-horizon agent when the worker dies mid-task, and what we built so the run survives it. - Per-user agent identity: we started on a shared service account and lost the ability to attribute what an agent did. Retrofitting identity was expensive. - Information boundaries: enforcing ethical walls and leaving a reviewable approval trail when an agent acts on a user's behalf. - Context engineering: the case where adding context back made agents measurably worse — a result no external benchmark would ever have surfaced for us. I will show the failures next to the wins, with real numbers and the denominators included, and demo the platform live. Every pattern here is architecture and engineering method, not a product walkthrough.

  3. 14:35

    Beyond the Pilot: Managing Your 500th AI Agent

    Abstract

    Getting your first AI application into production is hard; you hear that at every AI conference. But what happens when you're managing your 50th, 100th, or 500th agent? At enterprise scale, the challenges fundamentally change. It's no longer just about building agents that work. It's about managing hundreds of agents across teams, business processes, and systems while maintaining reliability, governance, visibility, and control. Who owns each agent? How do you monitor quality and performance? How do you prevent duplicated efforts, manage shared knowledge, and ensure that an expanding AI workforce remains trustworthy? Drawing on real-world enterprise deployments on Dify's AI Agent Management Platform, Andy Wang will explore what happens after AI agents successfully reach production. Topics such as the architectural decisions, operational challenges, and organizational lessons that rarely come up in conversations about pilots, but become impossible to ignore once you approach scale. The next frontier of enterprise AI isn't about how fast you build your first agent. It's about how effectively you manage your next 500.

  4. 14:40

    Break

  5. 14:55

    Beyond Reasoning: Building Infrastructure for AI Agents That Act

    Abstract

    AI agents are moving beyond reasoning and generation toward taking real-world actions on behalf of users. This shift requires a new infrastructure layer for identity, authorization, trust, credentials, and execution. Drawing on Visa’s work on Visa Intelligent Commerce (VIC), this session explores how these primitives are being designed for autonomous agents — how an agent proves who it is, demonstrates user intent, accesses credentials securely, and interacts with external systems safely. We’ll discuss the emerging infrastructure stack needed to move agents from assistants to trusted actors in the real world.

  6. 15:10

    Beyond Screenshots: Building Agent-Native Verification for Mobile Apps

    Abstract

    As one of the founding Android engineers on Instagram Edits, this talk shares my first-hand lessons from making Instagram Edits on Android agentic. Agentic coding has dramatically accelerated development output, but human verification cannot scale at the same rate. We built agentic verification to close that gap. Today, our agents run 2,000+ verification workflows per week across agentic QA, developer feedback loops, stress testing, and debugging. I’ll share how we made a complex mobile app discoverable, controllable, and verifiable by agents, and how that enables a broader agentic mobile development loop that improves both developer velocity and product quality.

  7. 15:20

    Agentic Harness: Real-Time Audio and Video AI Interviews

    Topic subject to speaker confirmation

  8. 15:30

    Break

  9. 15:40

    Your Repo. Your Benchmark.

    Abstract

    Turn thousands of your company’s merge requests into benchmarks to improve your coding agent harness.

  10. 15:50

    Trusted Context: Building the Layer Between Enterprise Data and AI Agents

    Abstract

    Most enterprise AI agent failures don't look like model failures. The model is fine. It just doesn't know which of four revenue tables finance actually uses, who owns it, whether last night's pipeline ran, or that one column was deprecated in March. That knowledge exists inside the organization - scattered across technical schemas, pipeline telemetry, business definitions, and org charts - and almost none of it reaches the agent. This talk is about assembling those four kinds of metadata into a single, model-agnostic context layer that agents can query, and what it actually takes to do that inside a large enterprise. I'll cover the architecture briefly, then spend most of the time on the parts that are harder than the architecture: deciding what context is trustworthy enough to serve, keeping it fresh, exposing it to agents over MCP, and the organizational work of getting teams to maintain context they never see themselves using.

  11. 16:10

    The Harness Is the First Reviewer

    Abstract

    Over the past year we modernized ATLAS end to end, backend and frontend alike: the cohort design workbench the OHDSI medical research network has run for over a decade. The Java backend moved from Java 8 to Java 21 and Spring Boot 3: some 1,600 files, with a net 65,000 lines removed. The frontend is a ground-up rewrite in Vue 3 and TypeScript: 110k lines of source, gated on every pull request by coverage-enforced unit suites and an end-to-end fidelity check against the entire OHDSI Phenotype Library, running publicly at atlas-preview.ohdsi.org/atlas. The large majority of that code was written by coding agents, far more change than anyone could have read line by line. The platform and the verification harness are open source under the OHDSI GitHub org, end to end. On the frontend rewrite I directed, reviewed, and accepted every change. When acceptance cannot mean reading every line, verification has to live somewhere else. The harness became the first reviewer, and I will show what it is made of: specs as the unit of delegation, the legacy implementation as a behavioral oracle, E2E suites and type gates as the merge criterion, and screenshot-driven iteration in place of pixel opinions. I will walk one real feature end to end and show the failure modes the harness caught, the ones it did not, and what I would gate differently today. This talk is for engineers and leads who need to deliver more than their team can hand-write. You will leave knowing how to structure specs so agents can execute them, which verification gates can replace line review and which cannot, and how to tell whether your project fits this model (rewrites with a reference implementation) or does not. Live instance: atlas-preview.ohdsi.org/atlas. All code: github.com/OHDSI/Atlas3.

  12. 16:15

    Break

  13. 16:35

    Your Data Agent Needs More Than a Knowledge Graph -- From Semantic Layers to Executable Ontologies with Apache Ossie

    Abstract

    * Semantic Layer as Infrastructure for Data Agents * Apache Ossie: Starting from Metric Interchange * From Specification to Implementation * What Makes a Semantic Layer AI-Native? * Ontology on Top of Semantic Models * Datus + Dosi: Building Semantic Layers and Ontologies from Enterprise Context * An Open Semantic Layer for the Agentic Data Stack

  14. 16:45

    From Idea to Viral Launch: How I Built a 3D Little Prince Game with AI

    Abstract

    The Little Prince: Across the Stars is a 3D story game built by one person with AI. This talk follows the project from adapting a familiar story into explorable scenes, through iterating on code, 3D assets and visual polish, to its public launch. Sac shares what worked, what did not, and what the process means for people creating with AI.

  15. 16:55

    From Matching to Delivery: Scaling Creator Partnerships with AI

    Abstract

    Through a demo of AhaCreator’s brand and creator interfaces, Susan shows how AI advances collaborations from brief to content delivery. Examples include a customer exceeding US$500,000 in collaboration spend in a single month and selected trend-driven campaigns moving from idea to publication in three to five days. The session explores how responsibilities are divided between AI and people, and how actual prices, response times, and delivery records inform future creator selection and partnerships.

  16. 17:00

    Breakout close

Four sessions from Shanghai

All speakers
01 / 03An onstage group portrait, from the conference album.
02 / 03The first stop: registration.
03 / 03A search agent, component by component.

Before Shanghai

What people took home

Shanghai community wall
BETWEEN TALKS
“I realized I hadn’t taken a single photo with all the people I’d been talking to.”
Wey Gu 古思为 · X ↗Translated from Chinese
FROM THE AUDIENCE
“Still, it's the best talk I saw on day 1 of AIE…”
ZinkLu · X ↗On filming Ryo Lu’s talk from the audience
THE DAY AFTER
“Seeing people share what they learned and who they connected with has been so rewarding.”
Sophia X. H. · X ↗AI Engineer

Our sponsors & partners

Thank you to the sponsors and partners supporting AI Engineer Shanghai.

Supported by

Organizations helping make this event possible.

Media & community partners

Create a community poster →Discuss sponsorship