Overview
The annual ACM SIGKDD Conference on Knowledge Discovery and Data Mining is the premier international forum for data mining researchers and practitioners from academia, industry, and government to share ideas, research results and experiences.
Sponsorship Details
Workshops, talks, and tutorials
KDD 2026 Workshop on Agentic Software Engineering (SE 3.0) The Rise of AI Teammates
August 9
Website: Link
Keynote Speaker: Behrooz Omidvar-Tehrani
Time: 8:00 - 12:00 PM
Location: Samda D
Description: We are witnessing the emergence of Agentic Software Engineering (SE 3.0), where AI agents act as autonomous AI Teammates performing complex tasks such as coding, debugging, and testing. As AI Teammates generate a vast new category of digital artifacts, they introduce unique opportunities and challenges related to human-AI collaboration, trustworthiness, and economic impact.
This workshop serves as the premier forum for addressing these challenges, anchored by the launch of the AIDev dataset. Comprising over one million agentic pull requests generated by AI Teammates such as Claude Code, OpenAI Codex, and GitHub Copilot, AIDev provides the empirical evidence needed to understand the behaviors of AI Teammates. This workshop features insights from major industry players and academic pioneers, and aims to define a roadmap for a world where AI Teammates and human developers build the future together.
Keynote Speaker: Behrooz Omidvar-Tehrani
Time: 8:00 - 12:00 PM
Location: Samda D
Description: We are witnessing the emergence of Agentic Software Engineering (SE 3.0), where AI agents act as autonomous AI Teammates performing complex tasks such as coding, debugging, and testing. As AI Teammates generate a vast new category of digital artifacts, they introduce unique opportunities and challenges related to human-AI collaboration, trustworthiness, and economic impact.
This workshop serves as the premier forum for addressing these challenges, anchored by the launch of the AIDev dataset. Comprising over one million agentic pull requests generated by AI Teammates such as Claude Code, OpenAI Codex, and GitHub Copilot, AIDev provides the empirical evidence needed to understand the behaviors of AI Teammates. This workshop features insights from major industry players and academic pioneers, and aims to define a roadmap for a world where AI Teammates and human developers build the future together.
KDD 2026 Workshop on Evaluation and Trustworthiness of Agentic AI
August 9
Link: https://kdd-eval-workshop.github.io/agenticai-evaluation-kdd2026/
Location: Jeju, Korea
This workshop advances evaluation and trustworthiness methodologies for agentic AI systems across their full deployment lifecycle, with particular emphasis on real-time post-market monitoring, model evolution, and production governance. As autonomous agents increasingly perform multi-step reasoning, planning, and action in open-ended real-world settings, traditional pre-deployment benchmarks and static evaluation frameworks prove insufficient.
We address core challenges including stochastic agent behavior, absence of ground truth, evolving user contexts, API-driven model updates, and lack of standardized metrics and audit practices. This workshop aims to foster interdisciplinary collaboration by bringing together researchers, industry practitioners, and policymakers to develop advanced evaluation techniques and governance frameworks for agentic AI systems that can be safely and reliably deployed in production.
Location: Jeju, Korea
This workshop advances evaluation and trustworthiness methodologies for agentic AI systems across their full deployment lifecycle, with particular emphasis on real-time post-market monitoring, model evolution, and production governance. As autonomous agents increasingly perform multi-step reasoning, planning, and action in open-ended real-world settings, traditional pre-deployment benchmarks and static evaluation frameworks prove insufficient.
We address core challenges including stochastic agent behavior, absence of ground truth, evolving user contexts, API-driven model updates, and lack of standardized metrics and audit practices. This workshop aims to foster interdisciplinary collaboration by bringing together researchers, industry practitioners, and policymakers to develop advanced evaluation techniques and governance frameworks for agentic AI systems that can be safely and reliably deployed in production.
KDD 2026 Workshop on Two-sided Marketplace Optimization
August 9
The TSMO 2026 (Two-sided Marketplace Optimization: Search, Discovery, Pricing & Growth) workshop will be held in conjunction with the KDD 2026 Conference. It is scheduled for August 9–10, 2026, as part of the official KDD 2026 Workshops program.
KDD 2026 Workshop on AI for Fraud and Abuse
August 9 - August 10
The 1st Workshop on AI for Fraud and Abuse (AI-FAB 2026) will take place at the ACM SIGKDD 2026 Conference on August 9–10, 2026, in Jeju, Korea. It focuses on tackling automated threats, synthetic identities, deepfakes, and balancing detection with economic realities.
KDD 2026 Workshop on Two-Sided Market Orientation
August 9 - August 13
Link: https://sites.google.com/view/tsmo2026/home
The 5th Workshop on Two-Sided Marketplace Optimization (TSMO 2026) focuses on multi-sided platform science, covering search, discovery, pricing, matching, and generative AI growth strategies. It takes place at the ACM SIGKDD 2026 conference in Jeju, South Korea (August 9–13, 2026).
The 5th Workshop on Two-Sided Marketplace Optimization (TSMO 2026) focuses on multi-sided platform science, covering search, discovery, pricing, matching, and generative AI growth strategies. It takes place at the ACM SIGKDD 2026 conference in Jeju, South Korea (August 9–13, 2026).
KDD 2026 Tutorial on Teaching LLMs to Write System Kernels for AI Accelerators: Post-Training, Reasoning, and Agentic Optimization
August 9
Website: Link
Authors: Youngsuk Park, Rajarshi Saha, Kaan Ozkara, Shaowei Zhu, Jiin Woo, Lingfan Yu, Wei Tang, Ziyang Xu, Zhen Jia, Emily Webber, Yida Wang
Time: Sunday, August 9th at 1:00 - 5:00PM
Location: Room 402A
Description: Writing high-performance system kernels for AI accelerators requires deep hardware expertise and low-level programming skill—a significant barrier for many researchers and engineers. This tutorial presents emerging methods that leverage large language models (LLMs) to automate kernel generation for GPUs and custom accelerators such as AWS Trainium. We cover the full pipeline: supervised fine-tuning on kernel corpora, reinforcement learning-based post-training (PPO, GRPO) with hardware-aware rewards, inference-time scaling for code reasoning, and multi-turn agentic systems for iterative kernel refinement. We also discuss key challenges unique to this domain, including reward hacking in compilation metrics, designing effective evaluation benchmarks for generated kernels, and strategies for data curation across heterogeneous hardware targets. The tutorial bridges the gap between LLM post-training research and systems programming, offering both foundational understanding and practical experience. We demonstrate how recent advances in reinforcement learning and inference-time compute scaling can be adapted to produce correct and performant kernels across diverse hardware backends. Attendees will gain hands-on experience generating Triton and NKI kernels using agentic LLM systems.
Authors: Youngsuk Park, Rajarshi Saha, Kaan Ozkara, Shaowei Zhu, Jiin Woo, Lingfan Yu, Wei Tang, Ziyang Xu, Zhen Jia, Emily Webber, Yida Wang
Time: Sunday, August 9th at 1:00 - 5:00PM
Location: Room 402A
Description: Writing high-performance system kernels for AI accelerators requires deep hardware expertise and low-level programming skill—a significant barrier for many researchers and engineers. This tutorial presents emerging methods that leverage large language models (LLMs) to automate kernel generation for GPUs and custom accelerators such as AWS Trainium. We cover the full pipeline: supervised fine-tuning on kernel corpora, reinforcement learning-based post-training (PPO, GRPO) with hardware-aware rewards, inference-time scaling for code reasoning, and multi-turn agentic systems for iterative kernel refinement. We also discuss key challenges unique to this domain, including reward hacking in compilation metrics, designing effective evaluation benchmarks for generated kernels, and strategies for data curation across heterogeneous hardware targets. The tutorial bridges the gap between LLM post-training research and systems programming, offering both foundational understanding and practical experience. We demonstrate how recent advances in reinforcement learning and inference-time compute scaling can be adapted to produce correct and performant kernels across diverse hardware backends. Attendees will gain hands-on experience generating Triton and NKI kernels using agentic LLM systems.
KDD 2026 Tutorial on Multi-Turn Reinforcement Learning for Large Language Models: From Theory to Practice with Amazon SageMaker AI
August 10
Website: Link
Authors: Sapana Chaudhary, Zhe Wang, Jiayu Li, Yuyang (Bernie) Wang, Xuan Zhu
Time: Monday, August 10th at 1:00 - 5:00PM
Location: Halla A
Description: Production LLM-based systems such as coding agents, web navigators, and tool-calling assistants operate over multiple turns of interaction with users, tools, and environments. This hands-on tutorial provides both a rigorous algorithmic and practical introduction to multi-turn RL finetuning for LLMs. Using Amazon SageMaker AI, participants progress through four labs: (1) environment and reward function design, (2) multi-turn trajectory collection and GRPO-based training, (3) reward densification and credit assignment strategies, and (4) evaluation, failure diagnosis and deployment.
Authors: Sapana Chaudhary, Zhe Wang, Jiayu Li, Yuyang (Bernie) Wang, Xuan Zhu
Time: Monday, August 10th at 1:00 - 5:00PM
Location: Halla A
Description: Production LLM-based systems such as coding agents, web navigators, and tool-calling assistants operate over multiple turns of interaction with users, tools, and environments. This hands-on tutorial provides both a rigorous algorithmic and practical introduction to multi-turn RL finetuning for LLMs. Using Amazon SageMaker AI, participants progress through four labs: (1) environment and reward function design, (2) multi-turn trajectory collection and GRPO-based training, (3) reward densification and credit assignment strategies, and (4) evaluation, failure diagnosis and deployment.
KDD 2026 Sponsor Invited Talks
August 11 - August 12
Tuesday, August 11th at 3:30 – 4:00 PM
- Speaker: Ming Ye, VP Amazon Prime Video Personalization and Discovery
- Title: Beyond Recommendation: Building the AI Foundation for the Next Generation of Entertainment
- Description: The last decade of entertainment AI focused on building large-scale recommendation systems that rank content. The next decade will be defined by AI systems that deeply understand customers as people, reason about intent and context, and orchestrate personalized entertainment experiences across every surface. In this talk, we share Prime Video's vision and strategy for this transition. Prime Video operates a global entertainment marketplace, spanning subscription, rental, purchase, live sports, and channels across 240+ countries where discovery is not a single ranking problem but a multi-objective, multi-format, context-dependent challenge unlike any other in the industry. We discuss how we are building unified customer intelligence that captures who a customer is across time, domains, and contexts; how AI is moving from ranking titles to orchestrating complete discovery journeys — from homepage to conversational interfaces; and how continuously learning foundations turn research into production impact at global scale. Along the way we highlight open research challenges in lifelong preference modeling, cross-domain representation learning, multimodal reasoning, and agentic discovery that we believe will define the next frontier of entertainment AI.
- Speaker: Yuxiang Wang, Principal Applied Scientist
- Title: Beyond Ranking: Building Prime Video's Discovery Systems at Scale
- Abstract: Prime Video builds a personalized homepage in real time from a catalog spanning movies, series, live sports, and channels. This talk goes inside the models and infrastructure behind it. We cover long-sequence behavioral modeling that captures evolving customer intent, and content embeddings that map a large, heterogeneous catalog into a shared semantic space for retrieval and cold-start. We then move from understanding single titles to orchestrating the whole page: multi-objective response prediction and whole-page construction, converging into joint optimization. We close with where recommendation is heading: foundation models and shared representations reused across ranking and retrieval, a move away from tuning isolated models toward building systems that continuously understand, learn, and orchestrate personalized discovery.
KDD Day on AI Reasoning
August 12
Website: Link
Invited Speakers: Junjie Tang (Sr. Principal, AWS), Behrooz Omidvar-Therani (Senior Applied Scientist, AWS)
Organizers: Luke Huan (Sr. Principal Scientist, AWS AI), Vijay Lingam (Sr. Scientist, AWS AI)
Time: 10:00 - 5:00 PM
Location: Room 201B
Description: Recent advances in large language models and foundation models have enabled powerful reasoning capabilities. Modern reasoning systems combine structured planning, tool use, and iterative refinement to solve complex tasks across domains such as coding, mathematics, scientific discovery, and decision making. The KDD Day on AI Reasoning brings together researchers and practitioners from academia and industry to explore emerging techniques for building next-generation reasoning systems.
Invited Speakers: Junjie Tang (Sr. Principal, AWS), Behrooz Omidvar-Therani (Senior Applied Scientist, AWS)
Organizers: Luke Huan (Sr. Principal Scientist, AWS AI), Vijay Lingam (Sr. Scientist, AWS AI)
Time: 10:00 - 5:00 PM
Location: Room 201B
Description: Recent advances in large language models and foundation models have enabled powerful reasoning capabilities. Modern reasoning systems combine structured planning, tool use, and iterative refinement to solve complex tasks across domains such as coding, mathematics, scientific discovery, and decision making. The KDD Day on AI Reasoning brings together researchers and practitioners from academia and industry to explore emerging techniques for building next-generation reasoning systems.
Booth Schedule
Tuesday, August 11
August 11
Demos
1:30 - 2:00pm - "Driving Data Center Efficiency with Energy Modeling", presented by Philip Christie (LinkedIn)
AWS operates data centers across diverse climates with widely varying cooling designs. To improve energy and water efficiency across this fleet, we built a physics-based model that replicates each facility from thermodynamic first principles. It produces hourly power and water targets for every component, which we then compare against real metered sensor data to find where inefficiency lives. Unlike pure ML approaches, the model tells us not just that something is wrong, but which specific piece of equipment is responsible. We discuss how this has contributed to a global PUE of 1.14 and a 52% improvement in water efficiency since 2021.
Come chat with us
1:30 - 2:30pm - Agent Evaluation, AI, Benchmarking, Benchmarking and Robustness, MCP+Agentic AI, ML, Reinforcement Learning Post-Training, Robustness, Synthetic Datasets, Tool Definition Quality
2:30 - 3:30pm - Agent Evaluation, Agent Memory, AI, Benchmarking, Benchmarking and Robustness, Entity Resolution, Graph ML, MCP+Agentic AI, ML, Multimodal Learning for Hierarchical Address Understanding, Reinforcement Learning, Robustness, Synthetic Datasets, Tool Definition Quality
3:30 - 4:30pm - Entity Resolution, Graph ML, LLM for Recommendations, LLM Training and Efficiency, Multimodal Learning for Hierarchical Address Understanding, Recommender System, Retrieval
4:30 - 5:30pm - Agentic Workflows, Causal Analysis, Decision-Making Over Multiple Inputs, ML Classification Models
1:30 - 2:00pm - "Driving Data Center Efficiency with Energy Modeling", presented by Philip Christie (LinkedIn)
AWS operates data centers across diverse climates with widely varying cooling designs. To improve energy and water efficiency across this fleet, we built a physics-based model that replicates each facility from thermodynamic first principles. It produces hourly power and water targets for every component, which we then compare against real metered sensor data to find where inefficiency lives. Unlike pure ML approaches, the model tells us not just that something is wrong, but which specific piece of equipment is responsible. We discuss how this has contributed to a global PUE of 1.14 and a 52% improvement in water efficiency since 2021.
Come chat with us
1:30 - 2:30pm - Agent Evaluation, AI, Benchmarking, Benchmarking and Robustness, MCP+Agentic AI, ML, Reinforcement Learning Post-Training, Robustness, Synthetic Datasets, Tool Definition Quality
2:30 - 3:30pm - Agent Evaluation, Agent Memory, AI, Benchmarking, Benchmarking and Robustness, Entity Resolution, Graph ML, MCP+Agentic AI, ML, Multimodal Learning for Hierarchical Address Understanding, Reinforcement Learning, Robustness, Synthetic Datasets, Tool Definition Quality
3:30 - 4:30pm - Entity Resolution, Graph ML, LLM for Recommendations, LLM Training and Efficiency, Multimodal Learning for Hierarchical Address Understanding, Recommender System, Retrieval
4:30 - 5:30pm - Agentic Workflows, Causal Analysis, Decision-Making Over Multiple Inputs, ML Classification Models
Wednesday, August 12
August 12
Demos
1:30 - 2:00pm - "Hierarchical CoFormer Sequence Modeling for Prime Video Home-Page Recommendation", presented by Andy Dong (LinkedIn)
The Prime Video Personalization & Discovery team builds the recommendation system that connects customers with content they love. We built CoFormer, a sequence model that co-form each customer's interest representation through a hierarchical, event-level fusion of their viewing history, served via conditional retrieval based on carousel intent. CoFormer composes every event from role-grouped specific tokens, what the customer watched, the context it happened in, how deeply they engaged, and who they are. A transformer over these enriched tokens learns where a customer's interest is heading next, substantially outperforming flat sequence baselines like SASRec on retrieval quality and cold-start coverage while surfacing the right titles into the best positions on the page.
Come chat with us
9:30 - 10:30am - Agentic Workflows, Causal Analysis, Decision-Making Over Multiple Inputs, ML Classification Models
10:30 - 11:30am - Agent Memory, Reinforcement Learning, Reinforcement Learning Post-Training
1:30 - 2:30pm - Entity Resolution, Graph ML, Multimodal Learning for Hierarchical Address Understanding
2:30 - 3:30pm - Entity Resolution, Graph Databases, Graph Mining, Graph ML, Graph Processing, GraphRAG, Knowledge Graphs, Multimodal Learning for Hierarchical Address Understanding, RAG
3:30 - 4:30pm - Agent Memory, Graph Databases, Graph Mining, Graph Processing, GraphRAG, Knowledge Graphs, RAG, Reinforcement Learning
1:30 - 2:00pm - "Hierarchical CoFormer Sequence Modeling for Prime Video Home-Page Recommendation", presented by Andy Dong (LinkedIn)
The Prime Video Personalization & Discovery team builds the recommendation system that connects customers with content they love. We built CoFormer, a sequence model that co-form each customer's interest representation through a hierarchical, event-level fusion of their viewing history, served via conditional retrieval based on carousel intent. CoFormer composes every event from role-grouped specific tokens, what the customer watched, the context it happened in, how deeply they engaged, and who they are. A transformer over these enriched tokens learns where a customer's interest is heading next, substantially outperforming flat sequence baselines like SASRec on retrieval quality and cold-start coverage while surfacing the right titles into the best positions on the page.
Come chat with us
9:30 - 10:30am - Agentic Workflows, Causal Analysis, Decision-Making Over Multiple Inputs, ML Classification Models
10:30 - 11:30am - Agent Memory, Reinforcement Learning, Reinforcement Learning Post-Training
1:30 - 2:30pm - Entity Resolution, Graph ML, Multimodal Learning for Hierarchical Address Understanding
2:30 - 3:30pm - Entity Resolution, Graph Databases, Graph Mining, Graph ML, Graph Processing, GraphRAG, Knowledge Graphs, Multimodal Learning for Hierarchical Address Understanding, RAG
3:30 - 4:30pm - Agent Memory, Graph Databases, Graph Mining, Graph Processing, GraphRAG, Knowledge Graphs, RAG, Reinforcement Learning
Thursday, August 13
August 13
Come chat with us
9:30 - 10:30am - Agent Memory, Agentic AI, Computer Vision, GenAI, LLM, Multimodal Representation, Reinforcement Learning, VLM
10:30 - 11:30am - Agentic AI, Computer Vision, Entity Resolution, GenAI, Graph ML, LLM, Multimodal Learning for Hierarchical Address Understanding, Multimodal Representation, VLM
9:30 - 10:30am - Agent Memory, Agentic AI, Computer Vision, GenAI, LLM, Multimodal Representation, Reinforcement Learning, VLM
10:30 - 11:30am - Agentic AI, Computer Vision, Entity Resolution, GenAI, Graph ML, LLM, Multimodal Learning for Hierarchical Address Understanding, Multimodal Representation, VLM
