GOSIM · SHENZHEN 2026
Schedule
Compare sessions happening at the same time and choose where to go next.
Filter tracksAll tracks
– 09:25
– 09:40
– 09:55
– 10:20
– 10:30
10:15–10:50 In progress
How Do You Read a Million Tokens on Your Laptop? You Don't.
10:15–10:50 In progress
openvela: From an Open-Source OS to the Foundation for AI Hardware Agents
10:15–10:50 In progress
Rethinking Residual Connections: Toward Depth-Effective Language Model Scaling
– 10:40
10:15–10:50 In progress
How Do You Read a Million Tokens on Your Laptop? You Don't.
10:15–10:50 In progress
openvela: From an Open-Source OS to the Foundation for AI Hardware Agents
10:15–10:50 In progress
Rethinking Residual Connections: Toward Depth-Effective Language Model Scaling
– 10:45
10:15–10:50 In progress
How Do You Read a Million Tokens on Your Laptop? You Don't.
10:15–10:50 In progress
openvela: From an Open-Source OS to the Foundation for AI Hardware Agents
10:15–10:50 In progress
Rethinking Residual Connections: Toward Depth-Effective Language Model Scaling
10:30–10:45 In progress
A DORA-Based 3D Navigation Framework: From LiDAR Point Clouds to Trajectory Planning
– 10:50
10:15–10:50 In progress
How Do You Read a Million Tokens on Your Laptop? You Don't.
10:15–10:50 In progress
openvela: From an Open-Source OS to the Foundation for AI Hardware Agents
10:15–10:50 In progress
Rethinking Residual Connections: Toward Depth-Effective Language Model Scaling
– 11:00
– 11:05
10:50–11:25 In progress
PenguinHarness: Efficient Self-Evolving Agent Harness for 1000+ Models
10:50–11:25 In progress
Building and Evolution of the Ascend Open-Source Training Ecosystem
– 11:10
10:50–11:25 In progress
PenguinHarness: Efficient Self-Evolving Agent Harness for 1000+ Models
10:50–11:25 In progress
Building and Evolution of the Ascend Open-Source Training Ecosystem
11:00–11:15 In progress
Trust Is Not the Default: From DORA-rs Dataflow Debugging to Trustworthy Tool Design
– 11:15
11:05–11:30 In progress
Agent-Centric Engineering — From Personal Productivity to Organizational Capability
10:50–11:25 In progress
PenguinHarness: Efficient Self-Evolving Agent Harness for 1000+ Models
10:50–11:25 In progress
Building and Evolution of the Ascend Open-Source Training Ecosystem
11:00–11:15 In progress
Trust Is Not the Default: From DORA-rs Dataflow Debugging to Trustworthy Tool Design
– 11:25
11:05–11:30 In progress
Agent-Centric Engineering — From Personal Productivity to Organizational Capability
10:50–11:25 In progress
PenguinHarness: Efficient Self-Evolving Agent Harness for 1000+ Models
10:50–11:25 In progress
Building and Evolution of the Ascend Open-Source Training Ecosystem
– 11:30
11:05–11:30 In progress
Agent-Centric Engineering — From Personal Productivity to Organizational Capability
– 11:35
11:25–12:00 In progress
From Connecting to Thinking: TuyaOpen's Agent-First OS for the Physical AI Era
– 11:45
11:25–12:00 In progress
From Connecting to Thinking: TuyaOpen's Agent-First OS for the Physical AI Era
– 12:00
11:25–12:00 In progress
From Connecting to Thinking: TuyaOpen's Agent-First OS for the Physical AI Era
– 12:15
– 12:30
– 13:30
– 14:00
– 14:25
– 14:35
14:00–14:35 In progress
OpenCV 5: What's New—and How to Prepare Your Company for the Upgrade
14:00–14:35 In progress
AI-Native Company Operating System: How Open Source Agents Transform Startup Workflows
14:00–14:35 In progress
Technological Innovation and Practice in the Database Ecosystem in the AI Era
14:00–14:35 In progress
Running Millions of Agents at Tencent: CubeSandbox as the Hands Behind the Harness Brain
– 14:50
– 15:10
14:35–15:10 In progress
SpecArtisans: A Verifiable Spec-Driven Development Paradigm for AI Coding
14:35–15:10 In progress
Project Homunculus: Open Source Force Feedback Glove for Humanoid Teleoperation
14:35–15:10 In progress
vLLM and Ascend: Plugin-Based Architecture and MoE Expert Load Balancing in Practice
– 15:15
14:50–15:15 In progress
Self-Evolving Apps: Building Software That Gets Smarter Every Time You Use It
– 15:40
15:10–15:45 In progress
RustNN at the Edge: WebNN as Infrastructure for Local AI Agents
15:10–15:45 In progress
From Hearing You to Seeing You — Real-Time Multimodality Brings AI into the Real World
15:10–15:45 In progress
From Vector Databases to the Data Foundation for AI Applications
– 15:45
15:10–15:45 In progress
RustNN at the Edge: WebNN as Infrastructure for Local AI Agents
15:10–15:45 In progress
From Hearing You to Seeing You — Real-Time Multimodality Brings AI into the Real World
15:10–15:45 In progress
From Vector Databases to the Data Foundation for AI Applications
– 16:00
– 16:05
15:40–16:05 In progress
Building a Knowledge-Work Harness: From Coding Agents to Executable Strategy
– 16:30
16:00–16:30 In progress
Building Hybrid AI Systems: Edge Inferencing Meets Cloud Reasoning
16:00–16:30 In progress
How to Use Harness Engineering to Build an AI Mini-Game Production Pipeline
16:00–16:30 In progress
Vastbase: AI Data Foundation for Intelligence-Driven Efficiency
16:00–16:30 In progress
Triton For Ascend：Simplifying High-performance Operator Development
– 16:55
– 17:00
16:30–17:00 In progress
Two Decouplings: Bringing Coding Agents into Everyday Applications
16:30–17:00 In progress
IntarkDB: The Underlying Data Infrastructure for Edge-Side Intelligence
– 17:20
– 17:30
17:00–17:30 In progress
From Prediction to Consequence: A New Evaluation Paradigm for Embodied AI
17:00–17:30 In progress
KunPeng SuperPoD - New Computing Infrastructure for the Agentic AI
– 09:25
– 09:40
– 09:55
– 10:40
10:15–11:15
Get started with Agent Development Kit (ADK)
*This is a hands-on session, please bring your own laptop.
– 10:45
10:15–10:45 In progress
Agentic AI on Global Edge: Deploying SPEAR-Powered Operations Agents with MCP
10:15–10:50 In progress
Developers' AI Creations: Co-building AI Hardware through the Lens of a Tech Community
10:15–10:50 In progress
Teaching AI to Improve AI: From Experimental Feedback to Meta-Evolution
10:15–10:45 In progress
"World Model for All": Towards Unified Generation, Navigation and Control
10:15–11:15 In progress
Get started with Agent Development Kit (ADK)
*This is a hands-on session, please bring your own laptop.
10:15–10:50 In progress
High-Quality Token Service: From Heterogeneous PD to Cross-Region AI Infrastructure
– 10:50
10:15–10:50 In progress
Developers' AI Creations: Co-building AI Hardware through the Lens of a Tech Community
10:15–10:50 In progress
Teaching AI to Improve AI: From Experimental Feedback to Meta-Evolution
10:15–11:15 In progress
Get started with Agent Development Kit (ADK)
*This is a hands-on session, please bring your own laptop.
10:15–10:50 In progress
High-Quality Token Service: From Heterogeneous PD to Cross-Region AI Infrastructure
– 11:05
10:15–11:15 In progress
Get started with Agent Development Kit (ADK)
*This is a hands-on session, please bring your own laptop.
– 11:15
10:50–11:25 In progress
The World as the Interface: Building an Open Agent Ecosystem for AI Glasses
10:15–11:15 In progress
Get started with Agent Development Kit (ADK)
*This is a hands-on session, please bring your own laptop.
– 11:25
– 11:45
11:15–11:45 In progress
From Khronos AI Ecosystem Research Insights to Edge AI Inference with Vulkan
11:25–12:25
Build multi-agent systems with ADK
*This is a hands-on session, please bring your own laptop.
– 12:00
11:25–12:00 In progress
MemOS: The Next-Generation Memory System for Autonomous Agent Evolution
11:25–12:00 In progress
Building AI Agent Devices on RISC-V with DC-ROMA Mainboard III
11:25–12:25 In progress
Build multi-agent systems with ADK
*This is a hands-on session, please bring your own laptop.
– 12:15
– 12:25
12:00–14:00 In progress
Lunch
12:15–14:00
Lunch
12:00–14:00 In progress
Lunch
12:00–14:00 In progress
Lunch
12:00–14:00 In progress
Lunch
12:15–14:00
Lunch
11:25–12:25 In progress
Build multi-agent systems with ADK
*This is a hands-on session, please bring your own laptop.
12:00–14:00 In progress
Lunch
– 12:30
– 13:10
– 13:30
– 13:40
– 13:50
– 14:00
12:00–14:00 In progress
Lunch
12:15–14:00 In progress
Lunch
12:00–14:00 In progress
Lunch
12:00–14:00 In progress
Lunch
12:00–14:00 In progress
Lunch
12:15–14:00 In progress
Lunch
12:00–14:00 In progress
Lunch
– 14:20
13:50–14:20 In progress
vLLM Latest Update and Roadmap
Check-in: 13:10–13:40. The workshop begins at 13:40.
– 14:30
14:00–14:30 In progress
Evaluation You Can Audit: Eclipse PanEval as Shared Infrastructure
14:00–14:35 In progress
AI-Native Embedded Product Development: Reshaping the Hardware-Software R&D Paradigm
14:20–14:50
llm-d: Building a Cross-Workload, Cross-Modal, Cross-Hardware, and Cross-Platform Distributed Inference Control Plane
Check-in: 13:10–13:40. The workshop begins at 13:40.
14:00–14:35 In progress
UCM: Beyond KVCache Acceleration, Lifecycle Management for Agent-Scale Inference
– 14:35
14:00–14:35 In progress
AI-Native Embedded Product Development: Reshaping the Hardware-Software R&D Paradigm
14:20–14:50 In progress
llm-d: Building a Cross-Workload, Cross-Modal, Cross-Hardware, and Cross-Platform Distributed Inference Control Plane
Check-in: 13:10–13:40. The workshop begins at 13:40.
14:00–14:35 In progress
UCM: Beyond KVCache Acceleration, Lifecycle Management for Agent-Scale Inference
– 14:40
14:20–14:50 In progress
llm-d: Building a Cross-Workload, Cross-Modal, Cross-Hardware, and Cross-Platform Distributed Inference Control Plane
Check-in: 13:10–13:40. The workshop begins at 13:40.
– 14:50
14:35–15:10 In progress
Efficient Training of Large Language Models via Orthogonal Transformation
14:35–15:10 In progress
Rethinking Efficiency Bottleneck of Generative Robotics Manipulation
14:20–14:50 In progress
llm-d: Building a Cross-Workload, Cross-Modal, Cross-Hardware, and Cross-Platform Distributed Inference Control Plane
Check-in: 13:10–13:40. The workshop begins at 13:40.
– 15:00
14:35–15:10 In progress
Efficient Training of Large Language Models via Orthogonal Transformation
14:35–15:10 In progress
Rethinking Efficiency Bottleneck of Generative Robotics Manipulation
14:40–15:00 In progress
Full-stage Integrated Practice of Open-source Robots and AI Education
14:50–15:20
AFD Plugin: Attention–FFN Disaggregation in vLLM
Check-in: 13:10–13:40. The workshop begins at 13:40.
– 15:10
14:35–15:10 In progress
Efficient Training of Large Language Models via Orthogonal Transformation
14:35–15:10 In progress
Rethinking Efficiency Bottleneck of Generative Robotics Manipulation
14:50–15:20 In progress
AFD Plugin: Attention–FFN Disaggregation in vLLM
Check-in: 13:10–13:40. The workshop begins at 13:40.
– 15:20
14:50–15:20 In progress
AFD Plugin: Attention–FFN Disaggregation in vLLM
Check-in: 13:10–13:40. The workshop begins at 13:40.
– 15:30
15:10–15:45 In progress
Interactive Heuristic Learning Method of LiDAR and Camera for Target Detection
– 15:40
15:10–15:45 In progress
Interactive Heuristic Learning Method of LiDAR and Camera for Target Detection
– 15:45
15:30–16:00 In progress
Cangjie and the New Frontier of Programming Languages in the AI Era
15:10–15:45 In progress
Interactive Heuristic Learning Method of LiDAR and Camera for Target Detection
15:40–16:00
Tea Break
15:40–16:10
From Single-Point Acceleration to System-Wide Optimization: Full-Stack Optimization Practices for LLM Inference
Check-in: 13:10–13:40. The workshop begins at 13:40.
– 16:00
15:30–16:00 In progress
Cangjie and the New Frontier of Programming Languages in the AI Era
15:45–16:00
Tea Break
15:45–16:00
Tea Break
15:45–16:00
Tea Break
15:45–16:00
Tea Break
15:40–16:00 In progress
Tea Break
15:40–16:10 In progress
From Single-Point Acceleration to System-Wide Optimization: Full-Stack Optimization Practices for LLM Inference
Check-in: 13:10–13:40. The workshop begins at 13:40.
15:45–16:00
Tea Break
– 16:10
15:40–16:10 In progress
From Single-Point Acceleration to System-Wide Optimization: Full-Stack Optimization Practices for LLM Inference
Check-in: 13:10–13:40. The workshop begins at 13:40.
– 16:20
16:00–16:30 In progress
The Agentic AI Developer Economy: Building a Meritocratic Open Source Ecosystem
16:00–16:30 In progress
An Autonomous Runtime Shielding Agent for Linux Kernel Vulnerabilities
16:00–16:30 In progress
AMD ROCm and Radeon-Powered Real2Sim2Real Vision-Language-Action Manipulation
16:10–16:40
Inside ROCm 10.0: Building an Open GPU Software Platform for the Agentic AI Era
Check-in: 13:10–13:40. The workshop begins at 13:40.
16:00–16:30 In progress
Turbo: Efficiently Serving Long-Context LLMs with In-Network Aggregation
– 16:30
16:00–16:30 In progress
The Agentic AI Developer Economy: Building a Meritocratic Open Source Ecosystem
16:00–16:30 In progress
An Autonomous Runtime Shielding Agent for Linux Kernel Vulnerabilities
16:00–16:30 In progress
AMD ROCm and Radeon-Powered Real2Sim2Real Vision-Language-Action Manipulation
16:10–16:40 In progress
Inside ROCm 10.0: Building an Open GPU Software Platform for the Agentic AI Era
Check-in: 13:10–13:40. The workshop begins at 13:40.
16:00–16:30 In progress
Turbo: Efficiently Serving Long-Context LLMs with In-Network Aggregation
– 16:40
16:20–16:40 In progress
From AI Access to AI Impact: Building Inclusive Education Across Cultures
16:10–16:40 In progress
Inside ROCm 10.0: Building an Open GPU Software Platform for the Agentic AI Era
Check-in: 13:10–13:40. The workshop begins at 13:40.
– 17:00
16:30–17:00 In progress
AI-Ready Data Production and Dynamic Data Training for Open-Source Models
16:30–17:00 In progress
PhyAgentOS: Recursively Self-Evolving Physical Agent Operating System
16:40–17:10
Scaling vLLM System-Wide: From Network I/O to Distributed Scheduling
Check-in: 13:10–13:40. The workshop begins at 13:40.
16:30–17:00 In progress
Beyond Compression Ratio: Service-Aware KV Cache Compression for Distributed LLM Serving
– 17:10
16:40–17:10 In progress
Scaling vLLM System-Wide: From Network I/O to Distributed Scheduling
Check-in: 13:10–13:40. The workshop begins at 13:40.
– 17:30
17:00–17:30 In progress
The Sovereignty Layer: Why Open Source Is How the Global South Shapes AI
17:00–17:30 In progress
Full-Stack Embodied AI OS — Introduction to openEuler Intelligence Boom Robot (IB-Robot)
– 17:50
– 17:55
– 18:00
17:55–18:00
Closing Ceremony
Time TBA
Time TBA
Rust Instructor Training & Workshop
Invitation only. The time and internal workshop details were not published.
No sessions match this day, time, and track selection. Try changing the filters.
All times are China Standard Time (UTC+8). Each window includes ongoing sessions; sessions spanning windows appear in each one.
Schedule categories
Keynote Plenary
6 sessions3F – ADEN (full hall)
- 09:10–09:55
- 09:10–09:55
09:1009:25
Empowering a Connected Intelligent World: The Symbiotic Evolution of OpenHarmony and AI
09:2509:40
UN ODET Open Source Strategy and Planning
09:4009:55
Open-Sourcing CANN to Foster Fertile Ground for AI Talent Development and Innovation
Archived schedule for GOSIM Shenzhen 2026. All times are China Standard Time (UTC+8).
Thank you for joining our biggest GOSIM ever!
ADEN Hotel Shenzhen Nanshan
October 16–17, 2026
Over 2,500 attendees came together in Shenzhen for GOSIM Shenzhen 2026. Thank you to every attendee, speaker, volunteer, sponsor, and partner!